[6s] [Music]
[17s] [Music]
[36s] [Music]
[42s] so
[44s] [Music]
[56s] so
[59s] [Music]
[1m 9s] all right everybody
[1m 12s] have you with us thanks for joining us
[1m 14s] this evening
[1m 16s] i'm just going to bring tommy
[1m 19s] hey hey luke
[1m 23s] good to have you back
[1m 26s] hope you had a nice stream last week
[1m 28s] with nunu
[1m 30s] yeah i mean it was nice not to have the
[1m 32s] code that was uh
[1m 34s] that was nice yeah have you been
[1m 40s] fine uh was a great week and
[1m 44s] yeah i looking forward to adding some
[1m 47s] new tests
[1m 48s] and also happy
[1m 52s] to see what you've added to pests the
[1m 54s] last two weeks
[1m 56s] it always sounds like you would maintain
[1m 58s] pests but
[2m] yeah nunu is the official maintainer but
[2m 2s] you are the
[2m 4s] feature machine i'm the one that adds
[2m 7s] all the breaking things
[2m 8s] and then yeah who knows the one that
[2m 10s] goes and fixes them
[2m 12s] yeah sorry nina yeah um
[2m 16s] okay one second i'm gonna bring my
[2m 21s] my screening all right
[2m 25s] yeah so uh a couple of weeks since we
[2m 27s] last looked at
[2m 28s] uh the open door repository
[2m 31s] and we have actually gone through a
[2m 34s] couple of minor
[2m 36s] patches on um pest
[2m 40s] so i'm pretty sure last time we were on
[2m 42s] the
[2m 43s] 1.12 so since
[2m 47s] the last stream on open. we've had 1.13
[2m 50s] and 1.40
[2m 52s] so to go over the main features in 1.13
[2m 56s] uh freyk added the two being expectation
[3m] which is kind of the opposite of to
[3m 1s] contain so kind of just to
[3m 4s] throw in what that would look like you'd
[3m 6s] have something like
[3m 7s] expect one to be in one
[3m 10s] two three they're quite nice if you're
[3m 14s] dealing with an array of
[3m 15s] various values and you just want to
[3m 17s] check that what you're expecting on
[3m 20s] is somewhere in there you're not too
[3m 23s] not too worried about where it is inside
[3m 26s] of
[3m 26s] that array that's a very nice addition
[3m 29s] we also had the
[3m 30s] clean shot interruption i have checked
[3m 33s] my calendar and
[3m 35s] if the dates are right for the release
[3m 36s] we were on one
[3m 38s] eleven so one twelve three added
[3m 41s] yeah okay so uh 112 is
[3m 44s] uh nice there are a few fixes going on
[3m 47s] um and the force option as well
[3m 49s] so if you're creating tests using the
[3m 52s] php arts and past test command
[3m 54s] you can now do dash dash force and it
[3m 56s] will overwrite even if it already exists
[3m 58s] which is just a nice little feature to
[3m 59s] have if you want to start fresh on a
[4m 1s] file that's
[4m 1s] badly wrong um and also support for php
[4m 5s] unit 9.3.7 which is good
[4m 9s] um something that oliver and i have
[4m 12s] been working out in the background for
[4m 14s] quite a while is the output in php storm
[4m 17s] which is a little bit of a devil so
[4m 20s] obviously when you run past tests
[4m 24s] in uh in the console they always look
[4m 27s] really nice
[4m 28s] and you get this beautiful output but
[4m 30s] previously when you were using the
[4m 32s] built-in
[4m 33s] test runner in phpstorm things were a
[4m 34s] little bit ugly
[4m 36s] well that's kind of been cleaned up um a
[4m 38s] lot in
[4m 40s] um 1.13 so you can see now we do get
[4m 44s] much nicer output
[4m 45s] but you still get all of the the links
[4m 47s] and error messages and everything you'd
[4m 48s] expect
[4m 49s] so hard to make this break
[4m 53s] here we go you get all of the nice error
[4m 55s] messages that phpstorm offers
[4m 58s] you've got your diffing and everything
[5m] like that you just get
[5m 2s] output that's a lot more similar to the
[5m 4s] pest console output so hopefully you'll
[5m 6s] feel a little more
[5m 7s] a little more at home now if you're
[5m 9s] using pest in php storm
[5m 13s] um and then recently uh as of
[5m 16s] two days ago maybe are we on the fifth
[5m 19s] the fourth as of yesterday
[5m 20s] time flies as of yesterday uh
[5m 24s] we added the bound closure to data sets
[5m 27s] and there is actually a test that we can
[5m 30s] demo this on
[5m 31s] this finds users by email test
[5m 35s] so obviously this has got a data set
[5m 36s] we're running four tests
[5m 39s] um there seems to be a little bit of a
[5m 41s] problem with the output here
[5m 43s] one second let's do it this way so we'll
[5m 46s] remove any distraction
[5m 50s] there we go four tests uh this example
[5m 53s] test is coming from
[5m 54s] php unit because pest doesn't know
[5m 58s] how to handle uh the php unit in an only
[6m] situation
[6m 2s] but that's not a problem is this we're
[6m 3s] interested in um yeah so previously
[6m 6s] what we've had to do is basically
[6m 10s] just pass the email into with which is
[6m 13s] fine
[6m 14s] um but there is actually a way we can we
[6m 17s] can use the new bound data sets to clean
[6m 19s] that up
[6m 20s] so just to try and demo that and then
[6m 22s] perhaps discuss how we feel
[6m 26s] we can come down here and i'm just going
[6m 27s] to open up my
[6m 29s] open door testing database
[6m 32s] conditions which should allow this test
[6m 34s] now to actually
[6m 36s] run
[6m 40s] just imagine if
[6m 45s] somebody had forgot to launch progress
[6m 50s] all right let's launch that
[6m 53s] stop that
[7m] cool okay so run that again
[7m 3s] okay so we get the the four passes
[7m 6s] which is nice uh but yeah again we're
[7m 8s] just passing an email in here
[7m 10s] which means we have to build the user up
[7m 12s] using tap each time
[7m 14s] now it's very much up to you as to
[7m 15s] whether you do this but basically
[7m 17s] um now in pest you could pass a short
[7m 20s] closure here
[7m 22s] and after passion passing the short
[7m 24s] closure you have complete access to the
[7m 26s] cascade itself
[7m 28s] so again
[7m 31s] perhaps not the best example of where to
[7m 33s] use it but in this instance we could say
[7m 35s] this user
[7m 37s] i spell your name right gummy bear and
[7m 40s] then
[7m 41s] you can you could call update and you
[7m 43s] could pass the different emails in
[7m 46s] but we don't pass we don't pass the
[7m 48s] different emails to
[7m 50s] the user we pass the different events to
[7m 53s] the
[7m 53s] to the scope even better then so we can
[7m 56s] say user
[7m 58s] by email and just pass
[8m 1s] that email there in and so now in the
[8m 4s] expect method you're not going to
[8m 5s] receive an email you're going to receive
[8m 7s] an instance of the user
[8m 11s] and is the data set resolved after the
[8m 14s] tab
[8m 17s] yeah the data set is resolved
[8m 21s] yeah it's a good point actually yeah
[8m 23s] because the tap will happen
[8m 25s] first let me just run it see what
[8m 27s] happens
[8m 29s] now you will get an exception because
[8m 32s] argument 1 isn't a type of user
[8m 45s] yeah okay and have your composer updated
[8m 55s] you're on 1.14 might be a little quirk
[8m 59s] with higher order tests
[9m 2s] let's look into that basically okay
[9m 5s] let's do this in a slightly different
[9m 6s] way but i definitely see
[9m 10s] the usage of these bond
[9m 14s] data sets for example sometimes i had
[9m 18s] some situations that i
[9m 20s] had to call the route function in my
[9m 23s] data provider
[9m 24s] yes definitely um in fact we're in a
[9m 26s] layer of the lap here right
[9m 28s] so yeah let's have we got any
[9m 31s] routes in this application i definitely
[9m 34s] have somewhere
[9m 36s] so let's do route string
[9m 40s] out and then jump and die about
[9m 43s] run that test
[9m 53s] i've still got only in here
[9m 58s] there we go so yeah you can see we
[10m] actually receive open door.test which is
[10m 2s] the resolve route
[10m 5s] um but in a data provider and this works
[10m 7s] with
[10m 8s] shared data sets as well which is really
[10m 10s] nice so you can
[10m 12s] you can hide this away inside a shared
[10m 14s] data set somewhere but
[10m 16s] in that shared data set still have full
[10m 18s] access to the
[10m 20s] uh to the test case
[10m 23s] and something i was playing around with
[10m 25s] yesterday as well which
[10m 26s] was quite nice is i don't know if you've
[10m 28s] done it like in php uni have you ever
[10m 30s] put like a bunch of bass tests in a
[10m 32s] trait that you want to run
[10m 34s] on every form of
[10m 37s] a certain test for example
[10m 42s] um obviously in a functional world we
[10m 45s] have to think of a different way to do
[10m 46s] that
[10m 47s] because you can drop a trait in there
[10m 49s] but you can have to write the tests in
[10m 51s] php
[10m 52s] unit syntax then um which is not a
[10m 55s] problem with per se but
[10m 56s] if you and this php unit you don't have
[10m 59s] access to
[11m] bound data sets exactly yeah and
[11m 4s] uh just i came across
[11m 8s] sorry this problem has bit me
[11m 12s] so often in my in my ass
[11m 15s] so that you you really can't use
[11m 18s] anything related to lateral in a data
[11m 21s] set
[11m 22s] so no config no route no
[11m 25s] eloquent you always have to to pass uh
[11m 29s] so simple values and do celebrate stuff
[11m 33s] in your test case
[11m 34s] and yeah it works and it
[11m 37s] is what pest does under the hood isu but
[11m 41s] yeah it's exactly to do it your own yeah
[11m 44s] it's exactly the same so
[11m 46s] i mean what happens in in php unit is
[11m 48s] it's just because the data set is is
[11m 50s] being resolved before the test case
[11m 52s] is booted and obviously it's in the
[11m 55s] segment set up
[11m 56s] method of the test case that laravel
[11m 58s] gets all of its features in tests
[12m] so the data set fails because of that
[12m 3s] but with
[12m 4s] uh when you pass a closure to the within
[12m 6s] pest test will wait
[12m 8s] to expand that closure until the test
[12m 11s] case has finished its setup method
[12m 13s] so that's how this works um which is
[12m 16s] nice
[12m 18s] and yeah it allows you to access laravel
[12m 20s] features it allows you to do more
[12m 21s] complex setup
[12m 22s] the situation yesterday where i had like
[12m 24s] um
[12m 26s] i had something like this swap and i
[12m 29s] wanted to
[12m 30s] i had like three or four different uh
[12m 33s] instances in the container that i wanted
[12m 35s] to do the exact same tests against each
[12m 36s] time
[12m 38s] so let's say this was like example class
[12m 41s] and i wanted an instance of my class
[12m 45s] and then i wanted to do that like let's
[12m 47s] say three different times
[12m 49s] but this one was another class
[12m 54s] and this one was a third class
[12m 58s] well now this is going to be run three
[13m 1s] different times you get three different
[13m 3s] tests
[13m 4s] but each time you're going to have a
[13m 6s] different instance resolved into the
[13m 8s] container
[13m 11s] could you show how the output
[13m 16s] of a failed bound data set
[13m 19s] uh test looks like yeah so this is
[13m 22s] something that needs
[13m 23s] a little more um polishing output wise i
[13m 27s] think
[13m 27s] let me let me show you so if we just
[13m 30s] resolve like one two and three
[13m 36s] data and we'll say expect
[13m 39s] data to be false
[13m 45s] but you can see you get you do get the
[13m 47s] real output
[13m 48s] but when you're looking at what is in
[13m 50s] the width you're going to get the
[13m 52s] closure
[13m 53s] you're going to get the closure object
[13m 55s] um whereas if this
[13m 56s] is obviously a plain data set and you
[14m 2s] you get the output of the actual data in
[14m 4s] there okay
[14m 5s] so it's something that needs a little a
[14m 8s] little more looking
[14m 9s] in the future i think and and can a bond
[14m 12s] data set return
[14m 13s] multiple values so can you return an
[14m 15s] array which is
[14m 17s] automatically spread over
[14m 20s] the argument conversation i was having
[14m 22s] with frake yesterday
[14m 25s] is let's get rid of the extra ones at
[14m 27s] the moment this is going to be passed in
[14m 29s] as an array right so if i come in by the
[14m 32s] array
[14m 36s] paper would be a better call there
[14m 37s] wouldn't it if i dump it down that array
[14m 40s] you can see
[14m 42s] that it is is in an array format
[14m 45s] what i was talking about yesterday with
[14m 47s] frake is
[14m 48s] is looking up the the tight pins and the
[14m 51s] number of parameters that you've
[14m 52s] requested
[14m 54s] and then deciding whether it should be
[14m 56s] passed as an array or whether it should
[14m 57s] be spread across
[14m 59s] um so that's something that will
[15m 2s] probably be out in one of the very
[15m 3s] next versions of pest but for now yeah
[15m 7s] if you return an array from one of these
[15m 8s] functions it's going to return the data
[15m 10s] as an array
[15m 11s] okay you can read more about bound data
[15m 14s] sets in the pest docs
[15m 16s] get started data sets and then bound
[15m 19s] data sets and you can see some examples
[15m 21s] that we do there
[15m 24s] very cool all right so i think you're up
[15m 27s] to date on fast
[15m 28s] yeah one more cool test i order
[15m 32s] not pretty cool but super simple i'm
[15m 34s] doing with
[15m 35s] data sets and several apps that don't
[15m 38s] have any test yet
[15m 40s] is just passing every single
[15m 43s] get route as and data set entry and do a
[15m 47s] simple
[15m 48s] sys get of this pest and route
[15m 51s] and expect 200 for example yeah
[15m 55s] that's a really good use case isn't it
[15m 57s] yes this is a
[15m 58s] super simple test and i i bet that a lot
[16m 1s] of test driven developers would say
[16m 3s] this isn't a test but one thing you can
[16m 6s] ensure this is
[16m 7s] if you don't have any other tests there
[16m 9s] is that you at least can call
[16m 11s] every get url on your page
[16m 14s] yeah and you get a valid result so you
[16m 17s] don't get any 500
[16m 18s] 400 for or whatever and
[16m 22s] yeah this is already a pretty good
[16m 24s] coverage
[16m 26s] if you don't have anything because it
[16m 28s] doesn't give you a lot of detail
[16m 30s] what is failing but at least it tells
[16m 33s] you if something fails
[16m 35s] yeah exactly and yeah i think if you
[16m 37s] were if you were coming into a test
[16m 38s] suite
[16m 39s] that was a full-blown application and
[16m 41s] you were thinking where do i even start
[16m 42s] with this
[16m 44s] to be able to whip up a data set like
[16m 46s] that with all your roots in it
[16m 48s] which you could probably actually do by
[16m 50s] copying and pasting the web.php file and
[16m 52s] then using a little bit of
[16m 54s] ide magic to transform everything into
[16m 56s] an array
[16m 58s] uh very quickly you could have built up
[17m] a test suite that you can expand
[17m 2s] and as you say will at least tell you if
[17m 4s] everything is working
[17m 6s] yeah yeah very nice one idea for
[17m 9s] bond data sets is it possible to have
[17m 13s] the closure on
[17m 14s] top level so this function
[17m 17s] and return the whole data set and not
[17m 20s] every single entry
[17m 23s] and now return an array of one two three
[17m 26s] for example
[17m 27s] and it will run a test case for one
[17m 30s] and one test case for two and another
[17m 32s] one for three i think
[17m 34s] that's part of shared data sets
[17m 40s] yeah but but these aren't
[17m 43s] bound to to my application
[17m 46s] so one idea of one idea for example
[17m 49s] would be to be able
[17m 51s] oh wait go back wasn't it data set
[17m 54s] you have a function you can yield yeah
[17m 57s] yeah
[17m 58s] so this looks like what i want
[18m 1s] and the nice thing about that as well if
[18m 3s] you're not familiar obviously and
[18m 5s] now you'll be familiar with yield
[18m 6s] tumbler if you're listening and you're
[18m 7s] not familiar with yield is
[18m 9s] that's like lazily handled so if you had
[18m 11s] a data set with
[18m 13s] a thousand ten thousand however crazy
[18m 16s] you want to go you could be loading in
[18m 17s] from a file somewhere or from a
[18m 19s] from an api somewhere it won't consume
[18m 23s] memory for all of those entries it'll
[18m 25s] handle the entries one at a time
[18m 27s] which is nice
[18m 30s] so yeah um there you go question answer
[18m 34s] happy days perfect hello to uh can and
[18m 36s] nuno and marion
[18m 39s] nice to have you with us
[18m 43s] all right so i was looking at
[18m 46s] where we got to the other day or the
[18m 48s] other week
[18m 49s] and work on this job right updates user
[18m 52s] details from here
[18m 54s] and so going into the job itself which
[18m 58s] is this one here
[19m] i think what we've tested is basically
[19m 2s] that
[19m 5s] correct correct me if i'm wrong but i'm
[19m 6s] pretty sure what we've actually tested
[19m 8s] so far
[19m 8s] is down to the line here because
[19m 11s] nowhere in our test do we then say well
[19m 14s] what if they have extra emails
[19m 16s] at the user emails endpoint
[19m 20s] so i think we need another test
[19m 24s] that is able to mock this end point here
[19m 29s] i think one of the reasons is that this
[19m 31s] endpoint is the same for
[19m 33s] every user um
[19m 38s] because it's uh it only works for the
[19m 40s] authenticated user
[19m 42s] so it doesn't work this is nicely for
[19m 46s] my current http fake uh setup
[19m 49s] yeah because
[19m 57s] so we could make it a bit more complex
[20m] this
[20m 1s] fake setup for example uh
[20m 4s] if you if you check the the if condition
[20m 7s] above
[20m 7s] um in the inside update user details job
[20m 15s] exactly so updating the emails only
[20m 19s] works if the user has a github token
[20m 22s] and if it has a github token this github
[20m 25s] token is set as
[20m 26s] br access token on the request
[20m 30s] so one
[20m 33s] relatively simple idea would be to
[20m 38s] to just set the username as access token
[20m 41s] for example
[20m 42s] and use this access token
[20m 47s] as well to generate a path
[20m 51s] but for some moment i think
[20m 55s] we would do better adding completely new
[20m 59s] tests
[21m 1s] covering other jobs and other parts of
[21m 4s] the application
[21m 6s] yeah okay because
[21m 10s] this definitely is something interesting
[21m 12s] but i don't know
[21m 14s] yeah it will take some time and i don't
[21m 17s] know how
[21m 18s] how patient our audience is
[21m 21s] yeah because i think you'd have to the
[21m 23s] the other way you'd do it is obviously
[21m 26s] you'd have to like whitelist a series of
[21m 30s] end points or file structures that had
[21m 33s] one particular way of doing things and
[21m 35s] then
[21m 35s] anything that not wasn't in the white
[21m 37s] list would have to go
[21m 39s] uh via a different fixture set okay
[21m 42s] would also be an idea
[21m 45s] which one of these jobs deals with
[21m 47s] something a little
[21m 49s] less um authenticated than what we've
[21m 51s] got here
[21m 56s] um so
[22m 1s] yeah so this one would definitely work
[22m 4s] yeah
[22m 4s] i like that one okay let's go
[22m 8s] for php artisan test test and we run it
[22m 12s] in the jobs directory
[22m 14s] and it's called load user
[22m 18s] repositories pressed
[22m 25s] yeah it doesn't have that page
[22m 29s] so just look at how we've done this one
[22m 31s] here obviously we're pulling in the
[22m 33s] uh in the job at the top and then we
[22m 36s] have this
[22m 37s] dispatch sync which is actually going to
[22m 39s] really run the job so we're not faking
[22m 41s] the job or anything like that we
[22m 42s] actually do all of the work
[22m 44s] exactly and we also don't depend on
[22m 49s] on the queue driver defined in the php
[22m 52s] unit
[22m 53s] we just want to run it yeah
[22m 58s] um so that being that being the case
[23m 1s] then let's say
[23m 2s] it and i guess we probably want to say
[23m 4s] that it loads
[23m 6s] we have already got a repos
[23m 10s] repos.json for barry vdh right
[23m 13s] yep so
[23m 17s] it saves the
[23m 20s] users
[23m 24s] other dress
[23m 27s] i'll do for now right yeah
[23m 30s] and you look at this job at the same
[23m 33s] time we
[23m 34s] kind of glean a few things so we
[23m 36s] probably want to do this this at first
[23m 38s] anyway but we don't want gummy bear this
[23m 41s] time we want barry vdh right
[23m 44s] he has repos unless you have a
[23m 47s] repost.json file that you can pull for
[23m 50s] yourself you can you can copy it just
[23m 53s] open your browser and go to
[23m 55s] api.github.com
[24m 1s] slash users slash gummybeer
[24m 6s] repose
[24m 11s] exactly nice
[24m 14s] so all endpoints i'm using are
[24m 19s] public you don't need any authentication
[24m 23s] out of interest because i saw something
[24m 26s] on twitter that you
[24m 27s] you'd kind of um
[24m 31s] mentioned but i i didn't read it
[24m 35s] like well enough to decide whether it
[24m 36s] was about this or not are there like
[24m 38s] rate limits on the public api for github
[24m 40s] yeah so
[24m 44s] without authentication you have six
[24m 46s] requests per hour
[24m 48s] um which is enough for some playing
[24m 50s] around and testing and so on
[24m 52s] um like this one just getting the
[24m 56s] json off of a specific endpoint
[25m] if you have a token uh independent if
[25m 3s] it's an o
[25m 4s] alt access token or a personal access
[25m 6s] token or whatever
[25m 8s] you get 5000 requests per hour
[25m 12s] and this sounds a lot um
[25m 15s] and for all and don't get me wrong but
[25m 19s] all smallish repositories from php
[25m 22s] and laravel world and compared to some
[25m 25s] other cs or small
[25m 28s] right yeah this is really no problem but
[25m 31s] as soon as you get to
[25m 33s] index microsoft spotify shopify
[25m 36s] ctngs codaforge is one for example that
[25m 40s] also
[25m 41s] hits rate limits several times
[25m 44s] these are repositories getting in real
[25m 46s] problems we are
[25m 48s] most time organizations as a whole
[25m 50s] because the organizations own several
[25m 52s] hundreds of repositories
[25m 54s] so between 300 to 800 or something like
[25m 58s] that
[25m 59s] so already only indexing also
[26m 2s] repositories takes
[26m 3s] up to eight to nine requests and now you
[26m 6s] have to
[26m 7s] get the repositories for every single of
[26m 10s] these repositories
[26m 11s] so these are already 800 requests
[26m 14s] and now most of the repositories don't
[26m 18s] have only
[26m 18s] 500 contributors but like 2 000 3000
[26m 22s] and up to and yeah this quickly sums up
[26m 27s] yeah and there are also repositories
[26m 30s] with around 25 000 contributors for
[26m 33s] example
[26m 34s] and my idea was to replace
[26m 38s] getting the contributors because this
[26m 40s] consumes the most requests uh
[26m 43s] from of all by native git
[26m 46s] because i can't get the same information
[26m 48s] by running git log
[26m 50s] i can get all the names i can get all
[26m 53s] the email addresses
[26m 54s] and match them to my database and
[26m 59s] it worked for playing around like also
[27m 1s] the first
[27m 3s] mvp of opener itself but also sen
[27m 7s] you know which repositories and so on
[27m 10s] are kind of the trouble guys
[27m 15s] and he runs a script for ctn js for
[27m 18s] example and
[27m 19s] five minutes later you get the warning
[27m 21s] of your mac os machines that your
[27m 23s] disk runs out of space
[27m 26s] okay and you check your file system and
[27m 29s] see that the city ngs repository takes
[27m 32s] up to gigabytes
[27m 37s] and the repository itself is only 90.
[27m 40s] whatever gigabyte so
[27m 41s] let's say 20. um so all these 110
[27m 46s] gigabytes are
[27m 46s] history git history and
[27m 50s] if you check the city ngs repository you
[27m 52s] see that they have around 100
[27m 54s] 000 commits and they get around
[27m 57s] one to five commits per hour and
[28m 1s] yeah if you have some
[28m 4s] only one binary file uh probably changed
[28m 7s] like images for example
[28m 10s] so it's quickly sums up yeah yeah man
[28m 14s] i suppose that's one of those things
[28m 15s] that like when you first build the
[28m 17s] application as well you're not exactly
[28m 19s] thinking about
[28m 19s] because you're like well who's going to
[28m 22s] use the application you know
[28m 24s] yeah it's the first time i've seen wow
[28m 27s] we have microsoft uh
[28m 29s] on board i said crazy
[28m 32s] and said i checked my queue and was like
[28m 34s] yeah every zoom runs through
[28m 36s] and i checked the queue one hour later
[28m 39s] and have seen
[28m 40s] four failed shops and i said
[28m 47s] and yeah six failed jobs per day
[28m 52s] are now my new zero because these are
[28m 55s] the jobs
[28m 56s] i i don't get around without really
[29m] changing code
[29m 1s] these aren't failing because of some
[29m 3s] real exceptions i
[29m 4s] definitely know what happens because
[29m 6s] it's rate limit
[29m 7s] but it's something i don't have full
[29m 9s] control over
[29m 10s] and i've done a lot to work around this
[29m 13s] but
[29m 15s] yeah somewhere there's a limit and
[29m 18s] now i have fine i have to find new ways
[29m 21s] to get around it
[29m 23s] yeah and one idea is now to have this uh
[29m 26s] git lock service so a service that runs
[29m 29s] this git lock
[29m 30s] behind the scenes but yeah it seems like
[29m 34s] i have to
[29m 36s] do it two ways so at first check the
[29m 39s] size of the repository for example
[29m 41s] and also the count of contributors i
[29m 44s] don't know if i can't get it
[29m 46s] and decide on these informations which
[29m 50s] way i want to go so if i want to go the
[29m 52s] git lock
[29m 53s] or say api away
[29m 57s] sounds very entertaining i'm glad i'm
[29m 59s] only writing the test suite
[30m 1s] [Laughter]
[30m 3s] uh oh hey to uh steve and hey to
[30m 7s] frank and marion asks is that the
[30m 10s] default document thing
[30m 11s] what theme is this you make a good point
[30m 17s] uh color scheme railcast's copy
[30m 20s] this is so yeah i actually
[30m 24s] i've been using this quite a while now i
[30m 26s] actually do like it it's nice
[30m 28s] i haven't changed a theme in other than
[30m 32s] trying light theme for like
[30m 34s] two hours i haven't actually changed
[30m 36s] theme for about
[30m 37s] a year and a half now which i think is
[30m 39s] saying something if you haven't changed
[30m 40s] steam for a year and a half
[30m 42s] you must really like the theme yeah i've
[30m 44s] run the north theme
[30m 46s] uh before for around three years
[30m 49s] and now i switched to photon by
[30m 52s] brent and originally by mozilla
[30m 56s] and yeah i tried to convince you
[31m 1s] maybe yeah maybe yeah i was too busy
[31m 3s] writing teachers
[31m 5s] to uh to give a thorough triumph
[31m 10s] um i don't know how you feel about this
[31m 12s] and some people will say you don't need
[31m 13s] to do it
[31m 14s] i personally like to do it and this is a
[31m 15s] really easy like we're going to come
[31m 16s] back to this don't we
[31m 18s] but um it has the expected
[31m 22s] timeout i think this is like an easy
[31m 24s] test that you can write
[31m 26s] and and just check so
[31m 29s] all i would do is is load up a new
[31m 32s] load user repositories job obviously
[31m 35s] that needs a user so let's pass that in
[31m 41s] and all i'm going to do is check that
[31m 42s] this public timeout property
[31m 44s] is set to this
[31m 48s] value here right and
[31m 51s] the reason i i do this
[31m 54s] and again you can you can complain at me
[31m 57s] uh
[31m 58s] tom if you disagree
[32m 1s] the reason i do this is if some other
[32m 4s] developer comes along in the future
[32m 9s] can we try to equal that if some other
[32m 11s] developer comes along in the future
[32m 14s] and they change this timeout i at least
[32m 17s] want them to think
[32m 19s] at least one thing about it i won't
[32m 22s] accept the pr if they change the timeout
[32m 25s] well yeah exactly yeah
[32m 29s] i suppose this depends on like how how
[32m 32s] the business works and in open source
[32m 34s] things work a little bit differently
[32m 35s] obviously
[32m 36s] but let's say could you replace
[32m 40s] a sys user by simply new user
[32m 44s] i imagine we could actually and in that
[32m 47s] case i would move this test to
[32m 49s] unit test because this test doesn't need
[32m 52s] anything
[32m 53s] it doesn't need a database no http
[32m 55s] nothing
[32m 56s] it just creates an instance and checks
[32m 58s] the property so
[33m] all right that by the way in
[33m 4s] article in case you're wondering if
[33m 5s] you're using the test command you can
[33m 7s] just do
[33m 8s] unit same as you can do for creating
[33m 10s] normal tests with
[33m 11s] php item that will create a unit test
[33m 14s] version
[33m 15s] um so yeah tommy is absolutely right we
[33m 18s] can copy that
[33m 19s] paste it here and now this hasn't got to
[33m 22s] go through the
[33m 24s] container at all uh
[33m 27s] we're gonna get an error on
[33m 33s] huh there's something happening inside
[33m 36s] the
[33m 42s] get back check check the
[33m 45s] parent of the job
[33m 48s] of course
[33m 51s] yeah no also only copied and then at the
[33m 54s] job itself
[33m 59s] no i don't know what all these traits do
[34m 4s] if they possible but
[34m 7s] but surely they'd have to be like talked
[34m 10s] about in the constructor wouldn't they
[34m 12s] but it complains about the eloquent
[34m 14s] model oh it's a searchable trait on the
[34m 17s] model
[34m 17s] oh damn
[34m 20s] yes because it's doing it inside yeah
[34m 22s] okay in which case
[34m 26s] um no no um
[34m 32s] i think the searchable should be a
[34m 34s] possible to disable
[34m 36s] so here here's a go up to the boot
[34m 40s] boot searchable yeah uh
[34m 43s] is there any register searchable macros
[34m 58s] okay so model observer you open some
[35m 2s] model observer
[35m 4s] linked in the arrow output yeah
[35m 9s] what's this one here
[35m 14s] okay yeah
[35m 17s] oh that's that's a little uh
[35m 21s] frustrating i mean it is what it is
[35m 27s] that's why i hate people doing stuff in
[35m 30s] the construct
[35m 32s] yeah definitely um i i think it's
[35m 36s] it's it's tough because it's one of
[35m 37s] those things that
[35m 39s] um it's easier to do it there
[35m 43s] most of the time but
[35m 47s] it does affect things further down the
[35m 48s] line and certainly like using
[35m 50s] lara i i certainly tried to avoid using
[35m 52s] laravel features in the constructor
[35m 55s] um advice i jumped to be fair i jumped
[35m 58s] between
[36m] um full feature tests and then
[36m 3s] trying to make everything unit test to
[36m 4s] avoid having to boot laravel
[36m 6s] so so even i changed from from
[36m 9s] project to project so let's move the
[36m 12s] test back to feature
[36m 13s] and lift with it
[36m 17s] for now
[36m 21s] okay so yeah our first test passes
[36m 24s] and again i say some people would say
[36m 26s] you don't need to do that
[36m 28s] or like don't bother with that
[36m 30s] personally i like to do it because
[36m 32s] it gives me an extra layer of security
[36m 34s] if something changes
[36m 35s] accidentally or someone else reviews the
[36m 38s] pr and they miss it
[36m 40s] so it's a little bit of documentation in
[36m 42s] my opinion obviously we can make this
[36m 44s] higher order
[36m 44s] with unexpected
[36m 54s] and that would just need to be
[36m 57s] that now
[37m 3s] okay all right
[37m 6s] uh let's just check the chat
[37m 10s] steve's a light light themed fan
[37m 15s] is there a nice way to use the vilt
[37m 16s] stack what's built
[37m 19s] i think it's vgs inertia laurel
[37m 22s] taillight
[37m 26s] i don't see why not i mean it
[37m 29s] it wouldn't obviously test the view side
[37m 31s] of things
[37m 33s] i guess you'd just be testing the api
[37m 34s] endpoints with past
[37m 36s] yeah but i think that's what yeah
[37m 40s] this industry you don't have real apis
[37m 43s] this is a
[37m 44s] i think kind of problem steve is facing
[37m 48s] yeah you just hit in like
[37m 51s] does inertia ship with test helpers
[37m 56s] i don't know i haven't used it let us
[37m 59s] know steve if
[38m] if a nursery ships with just helpers it
[38m 2s] might be worth looking into
[38m 4s] uh adding support for it
[38m 7s] either in the past core or in the
[38m 8s] inertia core so
[38m 12s] let's have a look another one
[38m 16s] another way to test it would be was task
[38m 19s] yeah which actually got
[38m 23s] pr into laravel itself
[38m 28s] wouldn't it be nicer to put the timeout
[38m 31s] before to equal yeah you can do that
[38m 33s] certainly yeah
[38m 35s] you can do that
[38m 39s] and in that case could we just do name
[38m 42s] it like
[38m 44s] job is probably configured for example
[38m 46s] and also checks that the queue is
[38m 48s] set to github yeah absolutely so we just
[38m 52s] get uh or do you want to
[38m 53s] we might as well use carry on using this
[38m 55s] is that what you're saying yeah just
[38m 56s] chain it and
[38m 57s] change the name
[39m 5s] should be github
[39m 9s] does it give her a blower case or if you
[39m 10s] don't get over there
[39m 15s] days and was there anything else you
[39m 18s] wanted to check
[39m 21s] i don't know yeah opens the job and
[39m 24s] almost like i have to do my own work
[39m 28s] uh load user repositories yeah you've
[39m 31s] set the time out to we extend github job
[39m 33s] which is obviously
[39m 34s] the reason why we can check that
[39m 38s] we've set the timeout don't think we're
[39m 40s] doing anything else as far as config
[39m 42s] goes
[39m 44s] and i guess if you're checking that the
[39m 45s] queue is github we can
[39m 47s] infer that we're extending the github
[39m 49s] job
[39m 51s] okay let's check if the github job does
[39m 53s] something
[39m 54s] really important but
[39m 57s] no that's the the
[40m 1s] the override is this isn't it that's
[40m 3s] what we're actually like
[40m 5s] explicitly stating on our job itself
[40m 13s] uh you can just do everything that
[40m 14s] livewire offers you yeah so like
[40m 16s] the reason i am don't personally use
[40m 19s] inertia all that much i've used on
[40m 20s] like one project when it first came out
[40m 22s] but i'm quite a live wire
[40m 24s] fan just because it's in php and that's
[40m 26s] where i'm most comfortable
[40m 29s] um yeah livewire works really nicely
[40m 31s] testing wise with
[40m 34s] and uh dusk as well which break you know
[40m 37s] i was saying
[40m 37s] frank put pr in the other week to
[40m 40s] laravel
[40m 42s] and i think you can just do dash dash
[40m 43s] pest
[40m 45s] now when you run the dust command and it
[40m 48s] will run it with pest rather than
[40m 49s] running it
[40m 50s] with
[40m 56s] um okay so now we need to do the actual
[40m 58s] job here right
[41m] let's come back to the user repository
[41m 4s] this paginated method which is
[41m 6s] interesting obviously doing a little bit
[41m 7s] of magic under the hood
[41m 10s] um so what is is this literally just
[41m 12s] gonna going to
[41m 14s] uh go over the pages one after the other
[41m 17s] and exactly
[41m 18s] yeah i think for the most part we can
[41m 22s] ignore that because we're basically
[41m 23s] going to get the full response anyway
[41m 24s] right
[41m 29s] kind of so now we always only check
[41m 33s] page one and
[41m 40s] this is fine for testing the logic
[41m 43s] chained for the response
[41m 44s] but for sure it doesn't test the
[41m 48s] paginated
[41m 48s] logic yes definitely so
[41m 54s] testing uh testing is a paginated logic
[41m 59s] could be
[42m 1s] bit problematic because
[42m 4s] right now it depends on the count of
[42m 7s] results and the response
[42m 9s] which isn't great so yeah
[42m 13s] we would have to define a json with
[42m 16s] exactly 100 entries
[42m 18s] and also at the second page so for the
[42m 22s] moment i think it's totally fine to test
[42m 25s] only the first page yeah because it's
[42m 28s] more likely that the error is and
[42m 32s] it's a chained logic and not in the
[42m 34s] pagination itself
[42m 35s] yeah and to be fair paginated is on the
[42m 38s] github job itself we might be able to
[42m 40s] come up with a job that is
[42m 42s] like a double inside the test itself
[42m 44s] that just generates a piece of json
[42m 46s] with exactly x amount of entries
[42m 49s] exactly yeah and we'll we'll just test
[42m 51s] paginated once when we come to it
[42m 53s] in this case we are going to load user
[42m 56s] repositories dispatch sync
[42m 58s] that is going to load uh gummy beers
[43m 1s] repos
[43m 3s] which is this here
[43m 7s] and i don't think there's like a is
[43m 9s] there a quick way in php storm to see
[43m 10s] how many entries we've got here
[43m 12s] i don't know that there is um but let's
[43m 16s] actually have a look right so we should
[43m 17s] be able to say
[43m 19s] this user positions
[43m 24s] dump
[43m 35s] one sets
[43m 40s] not so much no i was gonna say i'm
[43m 43s] pretty sure you've got more than one
[43m 44s] reaper
[43m 45s] okay yeah let's go through
[43m 49s] and take a look at what we've got could
[43m 51s] you do this user refresh at first
[43m 55s] i i didn't know after uh after
[43m 57s] suggestion
[44m] yeah so if we do this user refresh
[44m 2s] repositories
[44m 5s] still one okay interesting
[44m 9s] okay let's just do um
[44m 12s] jump and die the response here
[44m 15s] look at what we've got so there are 30
[44m 17s] items in that array for sure
[44m 21s] and then for each item in that array
[44m 23s] we're doing this
[44m 24s] from github yeah yeah
[44m 27s] we would have to fake also some
[44m 30s] responses
[44m 32s] or no from github handles
[44m 35s] hurry yeah so we get the array
[44m 41s] but it could be said said a lot of
[44m 45s] these repositories don't have a license
[44m 47s] and
[44m 48s] uh so we have some restrictions
[44m 52s] so they aren't allowed to be private
[44m 54s] they don't uh
[44m 55s] should be archived or disabled they have
[44m 58s] to have a license
[45m] yeah and
[45m 7s] still we get okay more than one
[45m 10s] passing that far so then it's gonna
[45m 13s] check
[45m 14s] the organization you are a user not an
[45m 16s] organization
[45m 18s] so it's going to grab the owner the
[45m 20s] owner should be you right
[45m 31s] yeah yeah there is
[45m 38s] so what's there what's so if this is a
[45m 39s] fork of another repository which again
[45m 42s] i'm pretty sure most of them aren't so
[45m 44s] if we were to
[45m 45s] dump and die here we should get a bunch
[45m 46s] of results again
[45m 48s] okay okay
[45m 52s] so it seems at this point if i coming
[45m 54s] out a second
[45m 56s] yeah so what's happening here i imagine
[45m 59s] is that
[46m] most of those repos are forks
[46m 4s] yeah let's check it the first one is the
[46m 6s] github
[46m 7s] which is excluded by name
[46m 12s] next one is a fork fork
[46m 16s] fork for no
[46m 20s] not fork but without the license
[46m 23s] yeah yeah reset
[46m 27s] fork
[46m 31s] through 25 entries
[46m 36s] and obviously fork faults yeah so far is
[46m 39s] only five and about four of those five
[46m 40s] have no license
[46m 42s] i should possibly clean up my
[46m 44s] repositories
[46m 47s] okay out of interest then if we switch
[46m 49s] this to barry vdh
[46m 53s] i mean this is interesting though we've
[46m 54s] just seen the code work in action
[46m 57s] yeah uh
[47m] okay switching switching to bayer vdh
[47m 2s] return one as well
[47m 8s] okay look these are these are mocked
[47m 11s] right these these are fake so what i'm
[47m 13s] going to do
[47m 14s] is i'm going to find some fork is
[47m 18s] fault or fork is true and change some
[47m 20s] defaults
[47m 22s] change some cooler repos i want to own
[47m 24s] yeah data transfer is okay
[47m 27s] let's just let's just write the four
[47m 30s] defaults on all 25 of these and then i
[47m 33s] can go back and selectively turn them
[47m 35s] off when i want to
[47m 36s] no i i wouldn't change it for all no you
[47m 39s] want me to say i
[47m 39s] can throw that back okay have you got
[47m 41s] any repairs in here that you're like i
[47m 43s] wish i owned that
[47m 46s] just click some together i don't care
[47m 50s] but because why i wouldn't change all
[47m 54s] to fork false is because in that case we
[47m 57s] wouldn't test the logics here
[48m 3s] so let's see how many that gives us
[48m 6s] still one
[48m 8s] yeah do you say have a license so once
[48m 11s] you have changed so let's change the
[48m 13s] data providers
[48m 23s] i don't want to get rid of fork faults
[48m 27s] see that one's there they're looking at
[48m 28s] it for fog faults
[48m 31s] and it has a hazardous
[48m 34s] are the other conditions on that
[48m 38s] um
[48m 44s] sorry from github isn't it
[48m 47s] disabled
[48m 56s] or data is uh holding
[49m 4s] i know okay
[49m 9s] see that would indicate that this one
[49m 11s] here so this is fork fault
[49m 16s] and it's also you know what else i'm
[49m 18s] gonna do by the way just
[49m 21s] just to avoid
[49m 24s] i don't know whether this will actually
[49m 26s] even work with
[49m 28s] normal okay
[49m 32s] um
[49m 35s] that license is mit and that for
[49m 37s] defaults which would indicate that that
[49m 39s] is the one
[49m 40s] is working adrian had a good question
[49m 43s] have you changed the username in the
[49m 46s] test case back to gamibia
[49m 49s] i thought or is it stillberry
[49m 52s] um uh
[49m 55s] i think you're right you know yeah yeah
[50m] good one adrian good catch and you're
[50m 3s] now here
[50m 4s] now okay happy days right
[50m 8s] close that down so at the very least we
[50m 10s] could say
[50m 13s] expect that the repositories count
[50m 18s] and i would do it uh repositories at the
[50m 20s] method and in that case
[50m 22s] you could remove the refresh yeah
[50m 26s] so do the count database side
[50m 31s] yeah right
[50m 34s] so the question now is
[50m 38s] technically we know that works there is
[50m 41s] a bit of a reliance
[50m 42s] on our fixture because we're basically
[50m 45s] saying that if the fix
[50m 46s] if you were to go and manually change
[50m 48s] that picture again this would break
[50m 51s] so if i was to go and change another one
[50m 53s] to to
[50m 54s] um to fork faults
[50m 58s] that would obviously update to five and
[51m] then we'd have to update our test
[51m 2s] which isn't like amazing but at the same
[51m 5s] time
[51m 7s] i think it's something that we can we
[51m 9s] can focus on a little bit later
[51m 12s] i guess what i would want to do is i
[51m 14s] would want to check the details
[51m 16s] of at least one of those repositories
[51m 18s] would
[51m 19s] was what i would expect it to be yeah
[51m 23s] yeah so if we grab
[51m 26s] the first repository and i'll do that
[51m 29s] with database
[51m 33s] we'll grab the first instance of an item
[51m 35s] from the repository
[51m 37s] and i guess that should be coming in
[51m 39s] here
[51m 40s] looking for four faults
[51m 44s] and then going to our very first
[51m 46s] instance and checking that this has a
[51m 48s] license
[51m 53s] let's find the next one that does
[51m 56s] neither does that one
[52m 1s] the cool thing about this is we're
[52m 3s] actually testing the other logic
[52m 7s] even without explicitly testing that
[52m 9s] logic
[52m 10s] like if you think about it while we're
[52m 13s] writing this test we've
[52m 14s] all by doing what we've just done we've
[52m 16s] already checked that those instances
[52m 18s] further up
[52m 19s] the first two instances that didn't have
[52m 21s] a license and never loaded in as
[52m 23s] repositories
[52m 24s] i've not explicitly written a test for
[52m 25s] that yet
[52m 27s] but you get that logic in this test and
[52m 29s] if we were to change how things worked
[52m 32s] it would break the test which is really
[52m 33s] nice exactly
[52m 36s] completely free right we didn't have to
[52m 38s] code that you just get that for free
[52m 41s] so sometimes like just thinking about
[52m 43s] what does this test actually cover
[52m 45s] helps you to understand when you need to
[52m 48s] write another test
[52m 49s] and when you've already written enough
[52m 51s] tests
[52m 53s] okay what actually is inside a
[52m 55s] repository
[52m 56s] now luckily for me you are very good at
[52m 59s] writing these
[53m 1s] so if i get this repo here
[53m 6s] and we have a look at the name
[53m 12s] okay so i would expect the name
[53m 15s] to be and i'll just check that that is
[53m 18s] indeed
[53m 20s] that property is name so the name should
[53m 23s] be
[53m 23s] here like that right yeah
[53m 27s] okay well
[53m 30s] i know it's it's a full name and
[53m 34s] okay yeah that's still interesting
[53m 36s] though isn't it
[53m 39s] so where is this in comparison to this
[53m 41s] one
[53m 43s] and is there some sort of sort ah okay
[53m 45s] isn't there a sorting algorithm on the
[53m 47s] repository
[53m 48s] yeah it could be yeah
[53m 51s] okay well we can we can have a look at
[53m 53s] that can we um
[53m 57s] it's like a global scope maybe
[54m 1s] yeah if so it should be on
[54m 4s] top
[54m 8s] like a trade
[54m 11s] no it's a remarkable
[54m 16s] but let's see where the repository
[54m 20s] armor icon class php is
[54m 24s] just going to do a little bit of window
[54m 26s] section here so i don't lose my face
[54m 30s] so if we if you would have line numbers
[54m 36s] enabled
[54m 38s] uh i think it's appearance yet
[54m 43s] five eight two and one nine four
[54m 46s] right okay okay that is interesting
[54m 50s] that is interesting so at some at some
[54m 53s] stage either we've got this wrong
[54m 55s] and this air is actually
[54m 58s] not being imported and so how we could
[55m 1s] we could work that out actually
[55m 3s] is we could come back into our test it
[55m 6s] could be that it is imported but
[55m 8s] uh not as the first one
[55m 11s] yeah exactly but i guess why there must
[55m 14s] be some
[55m 15s] there must be some form of sorting going
[55m 17s] on if that is the case
[55m 18s] let's grab the repositories and that's
[55m 21s] just map
[55m 22s] and graphs you can call plug on the
[55m 24s] repositories
[55m 26s] yeah so it's also possible but
[55m 30s] i always like to do a database site
[55m 41s] yeah yeah definitely in there so that
[55m 44s] the only
[55m 45s] explanation at this point just looking
[55m 47s] through the
[55m 52s] see the other thing is i was just
[55m 53s] thinking is that alphabetical order but
[55m 55s] then r
[55m 55s] is after l and it's definitely not an
[55m 59s] alphabetical order
[56m] it's like opposite alphabetical order
[56m 2s] for the a's
[56m 5s] and then randomly for the b's okay so
[56m 8s] unless something really weird is going
[56m 10s] on with jason and code which i i doubt
[56m 11s] is the case
[56m 13s] but that's definitely some of the proofs
[56m 15s] that you shouldn't depend on the order
[56m 17s] of json items
[56m 18s] no exactly yeah definitely and i mean we
[56m 21s] could obviously go and
[56m 23s] um when we do this first instead of
[56m 25s] doing first we could do first where name
[56m 28s] and grab the the one that we actually
[56m 30s] know is called
[56m 31s] gummy bear just kind of also do some
[56m 34s] expect uh to be true for example or
[56m 38s] something like this and do some
[56m 40s] user repositories where name
[56m 44s] every single of these four names yeah
[56m 46s] you mean like as
[56m 48s] uh just loading them in one after the
[56m 50s] other yeah
[56m 51s] or do an expect on the line we already
[56m 53s] have so this is a user repositories plug
[56m 56s] name
[56m 57s] yeah it could be a user expect
[57m 3s] him out there and then you can sink this
[57m 6s] is a bracket too much here
[57m 9s] contain and then you could literally
[57m 13s] which had not deleted that now one
[57m 14s] second
[57m 18s] we can literally take the output here
[57m 21s] but i think it shouldn't be to contain
[57m 24s] but the two equal
[57m 27s] yeah because to contain only does a
[57m 30s] subset check
[57m 33s] that's true i i i was kind of are we are
[57m 35s] we still saying that the order here is
[57m 37s] important then
[57m 41s] i think the order of items is only
[57m 42s] important for
[57m 44s] to be and not for to equal
[57m 48s] could it be i think that it's just going
[57m 51s] to use the assert equals in php unit
[57m 54s] yeah well let's try it if we run that
[58m 4s] obviously that's going to fail because
[58m 5s] of this expectation that's coming out a
[58m 7s] second
[58m 8s] if we run that and then let's drop that
[58m 10s] in there
[58m 11s] yeah so that's okay
[58m 14s] that's basically going to check the
[58m 17s] actual array contents it doesn't have to
[58m 19s] be the same
[58m 20s] object if it was 2b
[58m 23s] it'd have to be exactly the same
[58m 24s] reference it doesn't have to be the same
[58m 26s] reference but it does have to
[58m 28s] basically have the same contents in the
[58m 29s] same order
[58m 31s] um to contain is
[58m 39s] well maybe to match array
[58m 43s] okay nice oh yeah
[58m 46s] then i'm pretty sure you could
[58m 50s] uh
[58m 53s] oh this is this is
[58m 58s] why wouldn't you contain work
[59m 2s] thank you for messing around with two
[59m 3s] contain recently am i doing something
[59m 5s] weird here
[59m 7s] because of the array keys i'm pretty
[59m 9s] sure
[59m 13s] i i thought you could put multiple
[59m 23s] we could also do it another way
[59m 27s] and use higher order expectations
[59m 30s] and removes the two array call
[59m 35s] yeah and
[59m 38s] now you can use the contains method of
[59m 41s] the collection
[59m 44s] so plug returns a collection
[59m 48s] yeah so we could do that and then just
[59m 51s] call to be true
[59m 53s] i'm not sure if this works but at least
[59m 56s] it definitely works if you do it for
[59m 57s] every single name
[60m 2s] so i think this doesn't work so yeah
[60m 7s] so you're saying to go over each
[60m 10s] yeah you could do it with a sequence or
[60m 13s] just copies it contains
[60m 15s] to be true four times
[60m 20s] i use each yeah but each doesn't work i
[60m 23s] think
[60m 24s] no because it needs to be the way it's
[60m 26s] almost
[60m 30s] hold in yeah we do want to be in
[60m 34s] but you want to do each each
[60m 38s] to be in
[60m 41s] but this also doesn't verify at the end
[60m 45s] in theory that all four names appear
[60m 49s] yes true yeah you would also want to say
[60m 52s] that each one is unique
[60m 55s] um so personally i would just copy
[60m 59s] the contains chained call
[61m 2s] with every single name and
[61m 5s] asserts that every single name is to be
[61m 8s] true
[61m 20s] buttons here we wouldn't the sequence be
[61m 24s] a possible solution here well the
[61m 27s] kind of but sequence is ordering is also
[61m 30s] order specific
[61m 32s] yeah but if we do a contains
[61m 36s] check on the collection so order isn't
[61m 38s] it isn't important
[61m 44s] sorry tom i'm not following them
[61m 48s] so i'm
[61m 52s] checking the sequence documentation and
[61m 57s] yeah so you're saying drop a sequence in
[62m 1s] here
[62m 2s] exactly and
[62m 10s] each item sequentially there
[62m 14s] yeah so it walks over every
[62m 18s] item
[62m 21s] yeah so if i did that it would be in the
[62m 24s] order that i
[62m 25s] passed that in which is fine we are
[62m 28s] going to assert the order but if we're
[62m 29s] not interested in the order and we're
[62m 31s] only
[62m 31s] interested in checking that the data is
[62m 33s] actually there
[62m 36s] then that doesn't all of a sudden that
[62m 38s] kind of
[62m 40s] we're enforcing order again i mean
[62m 42s] before
[62m 43s] if we are going to enforce order we can
[62m 44s] obviously just use to match array
[62m 49s] or even to equal and just drop the
[62m 51s] actual array in there
[62m 52s] wait wait wait wait adrian already said
[62m 56s] something
[62m 56s] is there and to equal cannot say well
[62m 59s] something like to equal
[63m] cannot canonicalize
[63m 4s] yeah right all right all right very nice
[63m 8s] adrian says that he uses asset equals
[63m 10s] come out in kalysing
[63m 12s] in php unit let's just back this up then
[63m 15s] to that array
[63m 16s] faster just to rebuild the array
[63m 18s] wouldn't it do
[63m 20s] equal connectorizing yeah
[63m 23s] no idea what it does but adrian said
[63m 26s] that should solve our problem
[63m 28s] and
[63m 34s] and now change the order like dot files
[63m 36s] to the top or whatever
[63m 40s] ah perfect what a map what a man
[63m 44s] i think you should uh call adrian on the
[63m 47s] next stream
[63m 49s] i could that was that was next level
[63m 51s] nice nice uh
[63m 53s] and very guessable naming past as well
[63m 56s] yeah um
[64m 1s] see the other thing is i would expect
[64m 3s] canonicalizing to be like
[64m 6s] in canon which would obviously in my in
[64m 9s] my mind
[64m 9s] say that it should be ordered
[64m 13s] thanks adrian awesome okay so
[64m 16s] uh we can get rid of that we now check
[64m 19s] that those
[64m 21s] actually do exist and seeing as we
[64m 24s] actually grab this now we can also
[64m 27s] say to have count of four here which
[64m 30s] means you can get rid of that
[64m 31s] we don't need that anymore no passes
[64m 37s] very good okay get rid of that now it's
[64m 40s] getting a little bit
[64m 41s] busy okay
[64m 45s] so
[64m 49s] when we actually build this out we set
[64m 51s] the name to the full name
[64m 53s] the description to the description um
[64m 57s] i wonder if there is a
[65m 2s] useful way or rather let's put it this
[65m 6s] way
[65m 7s] is there data here that you are
[65m 10s] interested in checking specifically
[65m 22s] i think with names we already tested the
[65m 25s] primary job of this q job
[65m 32s] because now we really get into
[65m 35s] what should be tested on the repository
[65m 39s] itself
[65m 41s] so in a from github
[65m 44s] test doing all these different
[65m 47s] combinations as a data set for example
[65m 50s] but i think this would be far too much
[65m 53s] for
[65m 54s] a test on the q job definitely
[65m 57s] yeah because the main job of the queue
[65m 59s] job is loads this
[66m 1s] json response put it into chunks and
[66m 4s] throw every single chunk against the
[66m 6s] repository from github
[66m 9s] and again though
[66m 13s] even without going into the depth of
[66m 15s] writing
[66m 16s] the lower level test for this repository
[66m 19s] job from github
[66m 22s] if you only wrote this test
[66m 26s] you already do actually validate
[66m 29s] without being specific but you do
[66m 31s] actually already validate
[66m 33s] all of this logic here you possibly only
[66m 36s] have four repositories out of the other
[66m 37s] side
[66m 39s] exactly
[66m 43s] like if you only get around to writing a
[66m 45s] single test in your application
[66m 47s] per job per like class
[66m 51s] you're still testing the majority of
[66m 52s] that class if you throw enough scenarios
[66m 55s] at it which is why
[66m 56s] like this real world data i know we've
[66m 58s] had to edit it slightly but this real
[67m] world data is really useful
[67m 2s] because it gives the test something
[67m 5s] more realistic to work with rather than
[67m 8s] some perfect world made up thing that we
[67m 10s] create
[67m 12s] says also what i'm recommending to
[67m 15s] people starting with testing start with
[67m 20s] with the highest level of tests so
[67m 23s] like this uh get your home page and
[67m 25s] expect a
[67m 26s] status code of 200 and do some html
[67m 30s] checks for example like uh contains your
[67m 32s] brand name or whatever you want uh
[67m 35s] that's it that's your first test and you
[67m 38s] already test
[67m 39s] half of your application because you a
[67m 41s] router you test that
[67m 42s] your controller works on the home page
[67m 45s] most times you have
[67m 46s] like five calls to different models
[67m 49s] because you want to get some counts and
[67m 52s] fhqs and whatever and
[67m 56s] with such kind of tests you cover so
[67m 58s] much
[67m 59s] and yeah all these
[68m 3s] lower tests like uh unit tests for
[68m 6s] example
[68m 7s] are more to to get more precise feedback
[68m 11s] of your test suite so
[68m 13s] instead of it fails to rent a home page
[68m 15s] it tells you
[68m 17s] this specific method fails yeah
[68m 20s] definitely definitely um
[68m 25s] okay so talking about the second half of
[68m 27s] that then the last
[68m 28s] the the last thing we'll do on the
[68m 31s] stream today
[68m 32s] is look at this from github method in
[68m 34s] the repository
[68m 36s] which is like diving a bit deeper and
[68m 39s] rather than saying well we're assuming
[68m 41s] that this works
[68m 43s] we can actually assert this that this
[68m 46s] works
[68m 46s] i think also we should just have a
[68m 48s] little celebration that we've got
[68m 50s] nearly an hour and 10 minutes in and my
[68m 52s] microphone has been okay
[68m 54s] perfect i think that's enough that you
[68m 56s] should like probably start drinking
[68m 58s] again or something it should be that
[68m 59s] exciting
[69m 3s] very nice
[69m 7s] i agree adrian we should definitely have
[69m 9s] an alias in past
[69m 10s] more obvious but yeah thank you very
[69m 12s] much for pointing out the uh
[69m 14s] the name saved up awesome okay
[69m 17s] let's fire a test a test name out then
[69m 20s] we're just tackling this very first part
[69m 23s] we know that if we call from github
[69m 26s] and the data looks like well
[69m 30s] the data has any of these which is
[69m 31s] perfect case for a data provider
[69m 34s] then we should return null yeah so
[69m 38s] we can say um the
[69m 42s] and it might actually be best to put
[69m 44s] this into a separate test case itself
[69m 46s] but
[69m 46s] we were abstract too early let's say the
[69m 49s] front github method
[69m 54s] burns early if the data
[69m 58s] is invited
[70m 4s] um so i'm going to receive
[70m 8s] let's think about the best way to do
[70m 9s] this right let's say we receive
[70m 11s] an array of overrides overrides
[70m 16s] and let's create a bunch of valid data
[70m 19s] up here
[70m 20s] so let's say validator
[70m 23s] is an array we know that if private
[70m 26s] is false if
[70m 30s] archived is false
[70m 34s] if disabled is false
[70m 39s] if the license
[70m 45s] and it has to be an area i think
[70m 48s] that is an array and in fact let's go
[70m 51s] one step further
[70m 53s] let's jump back into our repos
[70m 57s] so important one is uh spd
[71m 1s] x underscore id
[71m 6s] what's this one here the short name of
[71m 10s] the uh the thing right
[71m 13s] yeah uh which one of these would
[71m 16s] replace quotes yeah that wouldn't be it
[71m 19s] for all of them
[71m 21s] at the same time
[71m 25s] you can keep them uh hpcs fix it will
[71m 27s] fix it
[71m 30s] okay so the license is like that
[71m 35s] and run and the name
[71m 39s] is not github
[71m 42s] dot github and not the username
[71m 45s] let's put we'll actually put like gummy
[71m 48s] beer in there right
[71m 50s] no suspect problematic
[71m 54s] name it something like php minus library
[71m 57s] or whatever
[72m 4s] it's it's the last larger or condition
[72m 8s] if the owner of the repository is of
[72m 11s] type user
[72m 12s] and the name of the repository
[72m 16s] equals the name of the user
[72m 19s] the repository is also ignored yeah
[72m 22s] so we're going to create an array like
[72m 24s] this with the owner
[72m 26s] and we'll say type we will make it a
[72m 30s] user because it belongs to you
[72m 33s] but the owner login
[72m 38s] yeah the owner log in would be gummy
[72m 39s] bear right
[72m 41s] yep two ends one m
[72m 46s] no double um yeah but that
[72m 50s] but because the name isn't equal to
[72m 53s] that login name everything's fine okay
[72m 57s] yeah okay now so why
[73m] exclude c's is not set you
[73m 4s] new but relatively uh sees
[73m 8s] repositories for users allowing the
[73m 11s] users to have some cool profile
[73m 13s] information on their profile
[73m 16s] possibly you have seen it already if you
[73m 18s] check some user profiles they have this
[73m 21s] this read me block at the top of their
[73m 24s] profile as you can put whatever you want
[73m 26s] so it's just a markdown file
[73m 28s] and this has to be put into a repository
[73m 32s] named the same as your username
[73m 35s] and because this is definitely not
[73m 37s] something open source i just excluded
[73m 42s] yeah okay now it's just um
[73m 47s] of working with the github repositories
[73m 53s] okay so basically like
[73m 56s] the most basic check is if we array
[73m 58s] merge our valid data with our override
[74m] data which we're going to provide in a
[74m 1s] second
[74m 3s] we expect the value to be normal
[74m 7s] i think as well so that this reads nicer
[74m 9s] i might override as a
[74m 10s] radical argument and that'll become
[74m 13s] clear of the data provider
[74m 16s] um all right let's just let's just try
[74m 19s] this out
[74m 20s] so we want to say with and we'll data
[74m 23s] provider
[74m 24s] and then we want to start providing some
[74m 26s] data um
[74m 30s] now we're relying on php 8 here i'm
[74m 32s] pretty sure that this will still work
[74m 34s] with
[74m 34s] array keys but let's find out if not we
[74m 38s] might have to provide it as an array
[74m 39s] inside an array
[74m 42s] but if we were to say
[74m 46s] private is equal to true
[74m 49s] and just to see if this actually works
[74m 52s] right
[74m 55s] okay so it's reset that's fine in which
[74m 57s] case we can't do it in a veracratic way
[75m] we have to do an array of arrays
[75m 4s] there we go okay let's
[75m 7s] move that down a little bit now and i'll
[75m 10s] type
[75m 11s] that again
[75m 15s] test and we pass
[75m 19s] so the nice thing about this is we can
[75m 22s] we've obviously got the data that works
[75m 24s] we can change one thing
[75m 26s] at a time and we can have as many data
[75m 28s] providers as we want or as many
[75m 30s] sets of data as we want and we're just
[75m 32s] going to test for each
[75m 34s] different scenario and make sure that
[75m 36s] each scenario is actually going to work
[75m 38s] so if archived is equal to true for
[75m 40s] example
[75m 43s] uh if disabled is equal to true
[75m 49s] uh if the license is null so
[75m 52s] license will set to null
[75m 56s] um if the name
[76m] is equal to github
[76m 6s] and then this this more complex one
[76m 8s] which is still
[76m 9s] fairly straightforward we're going to
[76m 10s] set the name
[76m 12s] to gummy bear
[76m 16s] and then we're going to set the
[76m 19s] owner of an array
[76m 23s] where the type is user and
[76m 26s] the login is also gamibia
[76m 30s] correct yep
[76m 33s] okay very nice
[76m 36s] very nice so
[76m 40s] obviously with just the one test and the
[76m 42s] set of valid data
[76m 44s] we're able to check all of these
[76m 45s] conditions
[76m 47s] and if an an extra condition came up in
[76m 50s] the future
[76m 51s] adding to this test would just be
[76m 52s] dropping in another line in the data set
[76m 54s] and changing changing what's there so
[76m 57s] you can use this for
[76m 58s] loads of different scenarios i use this
[77m] for validating
[77m 2s] requests for example this is like how i
[77m 5s] like to approach
[77m 6s] checking all of the validation methods
[77m 9s] in a request
[77m 10s] i just pass it real data merge it with
[77m 12s] some data that's bad
[77m 13s] and then check that we actually get a
[77m 15s] single validation type or error
[77m 19s] the different ways to tackle that but
[77m 20s] that's how i like to do it
[77m 26s] i guess the other thing is obviously it
[77m 28s] returns null which is which is
[77m 30s] great but we also want to check that it
[77m 32s] never actually
[77m 33s] [Music]
[77m 36s] let's actually build the repository
[77m 41s] so repository is created i don't know if
[77m 44s] it's
[77m 44s] really at the end but yeah it's not
[77m 47s] create
[77m 49s] so i guess the the only other test i'm
[77m 52s] thinking we want to do is we just expect
[77m 54s] at the users repositories
[77m 58s] or the repository all
[78m 1s] the empty or you could do repository
[78m 2s] count
[78m 5s] zero in other words yes i know that it
[78m 8s] returned null but
[78m 10s] just make sure it didn't return no
[78m 11s] further down or something weird happen
[78m 14s] i think really readable would be
[78m 16s] repository exists
[78m 17s] to be false very nice yeah
[78m 21s] absolutely
[78m 25s] awesome so that test is obviously now
[78m 29s] pretty foolproof um again something you
[78m 32s] can do
[78m 32s] choose to do or choose not to do is you
[78m 35s] can actually drop this in
[78m 37s] like up here nothing stopping you doing
[78m 40s] that you'll need to
[78m 41s] use the valid data
[78m 44s] uh also if you're going to do that you
[78m 45s] should name space the
[78m 47s] class that's something i found out
[78m 48s] yesterday from oliver but the other
[78m 50s] thing you can do
[78m 51s] is we can introduce a function here
[78m 55s] where before each we set this as this
[78m 57s] valid data
[79m] and when we do that now we can access
[79m 3s] this from this
[79m 4s] added data which is fine and the reason
[79m 8s] i'm doing that is because it would allow
[79m 9s] us just to convert this to a higher
[79m 10s] order test if we wanted
[79m 12s] so we would just use this expect here
[79m 21s] and we'd obviously need to give this um
[79m 24s] uh
[79m 24s] would we yes you need to overwrite
[79m 28s] we need the data right overrides
[79m 33s] and we would expect repository exists
[79m 36s] which is obviously we don't need the
[79m 39s] overrides but we do want
[79m 41s] to be in laravel when we run it
[79m 45s] and we can run to defaults
[79m 48s] and you still want all of the data so
[79m 51s] we'll grab that
[79m 57s] okay um at this point
[80m] i would say that every test
[80m 5s] more or less every test in this um
[80m 8s] directory in this test case
[80m 9s] should probably be to do with this data
[80m 12s] in some way
[80m 13s] um again it's an argument that you could
[80m 16s] say yes or no to
[80m 18s] but i would probably if there were other
[80m 21s] things in the repository i wanted to
[80m 22s] test down the line
[80m 24s] i would probably end up creating a
[80m 25s] subdirectory and models called
[80m 27s] repository
[80m 28s] and then i would have for instance uh
[80m 31s] the test might
[80m 32s] j uh the test name might be changed to
[80m 33s] from github test
[80m 35s] yeah um i would probably go down that
[80m 38s] line just so that the tests are
[80m 40s] a little bit more segregated and
[80m 42s] everything that's happening in like
[80m 43s] before each for example
[80m 44s] is only interested in one particular
[80m 46s] thing
[80m 48s] one other thing you can definitely
[80m 50s] already add to validator is an id
[80m 53s] key okay yeah
[80m 57s] that's up here and what would that be
[81m 1s] set to
[81m 2s] as a valid one one two three four five
[81m 5s] seven
[81m 6s] so it's a long numeric id okay
[81m 11s] nothing special fine
[81m 14s] okay so obviously this this is this
[81m 17s] block completely handled
[81m 20s] the next thing is you're saying look if
[81m 22s] it's an organization
[81m 24s] then we need to grab the organization
[81m 28s] from github which we did in another test
[81m 31s] case right
[81m 33s] uh we haven't did it yet i think
[81m 40s] organization
[81m 43s] yeah yeah okay
[81m 46s] it tests it indirectly this is
[81m 49s] a cis organization method yeah
[81m 53s] oh if yeah if we're changing the type to
[81m 54s] organization
[81m 57s] um then it's going to grab the owner as
[82m 1s] as an organization let's just skip that
[82m 4s] one for now
[82m 5s] because this is the default from our
[82m 6s] valid data right
[82m 8s] yeah so by default we're going to get a
[82m 11s] user anyway based on the valid data
[82m 13s] i guess the other thing we could do
[82m 15s] right now is let's write a test that
[82m 17s] says it
[82m 18s] throws an invalid
[82m 21s] argument exception if the
[82m 25s] owner type is not
[82m 30s] user or organization
[82m 37s] um so in this case we would literally be
[82m 40s] probably tapping this and we just want
[82m 44s] to
[82m 45s] say repository from github
[82m 50s] and pass in array merge
[82m 53s] this
[82m 56s] valid data and then
[83m 1s] owner
[83m 5s] type and we'll make the type
[83m 8s] equal to ubar they're an actual valid
[83m 11s] type in github that's not a user or
[83m 13s] organization nope
[83m 17s] so we would expect that that throws
[83m 22s] invalid argument exception
[83m 27s] awesome yeah and you have a deeper
[83m 31s] its user or organization and not off
[83m 35s] oh yeah thank you we could actually even
[83m 38s] check that the
[83m 39s] message is correct as a second argument
[83m 44s] so that would be basically that right
[83m 48s] yeah and in square brackets
[83m 52s] it's a whopper
[83m 57s] still didn't like me did it i missed
[84m] something
[84m 3s] i'll try getting rid of that
[84m 10s] hates me for writing repository on a
[84m 14s] type
[84m 15s] rid of that and now it's now it's
[84m 19s] failing all the time
[84m 21s] i must have killed it with something
[84m 22s] else what have i done
[84m 29s] could it be that uh
[84m 33s] okay look at that that's weird that was
[84m 35s] off
[84m 36s] change it to war oh because you
[84m 40s] retry also php storm for the win
[84m 45s] okay let me let me revert that awesome
[84m 48s] so um yeah if you're doing the uh the
[84m 52s] retry on
[84m 54s] phpstorm and you have changed the test
[84m 57s] name it won't be able to find the test
[84m 58s] and it will fail
[85m] obviously brilliant so now we're
[85m 2s] actually checking the message as well
[85m 4s] which is nice
[85m 6s] uh okay so let's skip past
[85m 10s] that for now we'll come back and add
[85m 12s] something for organization in a moment
[85m 14s] if the owner is an instance of user and
[85m 16s] the data
[85m 18s] is a fork is that what we're saying
[85m 20s] there yeah
[85m 21s] so we can actually just add that to
[85m 26s] this right we can just say
[85m 32s] or true yeah
[85m 35s] and you should add the fork false to
[85m 38s] to the valid data yeah
[85m 50s] let me find the reiki id
[85m 58s] oh you have to define an id
[86m 2s] in the valid data for
[86m 19s] and adrian asked if it wasn't possible
[86m 22s] to assert for an exception at the end so
[86m 25s] if you could switch throws and tap
[86m 29s] you can indeed yeah you can indeed
[86m 33s] okay yeah which which is actually
[86m 36s] uh usually i've skipped over it thinking
[86m 39s] about other things but
[86m 40s] usually my preference because i prefer
[86m 43s] to have everything to do with
[86m 45s] expectations or checks at the end of the
[86m 47s] test
[86m 49s] preference it is to have it at the top
[86m 53s] because it reads
[86m 54s] better uh it throws
[86m 57s] invalid argument exception actually that
[87m] yeah that's
[87m] that's certainly a a very valid argument
[87m 3s] i i get that
[87m 4s] i i was always a little bit like
[87m 7s] obviously if you're writing a proper php
[87m 8s] unit test
[87m 9s] let's say that this is the php unit test
[87m 11s] you have to do this
[87m 12s] expect exception yeah and then the
[87m 16s] acceptance type
[87m 17s] and and every other test will have like
[87m 19s] this
[87m 20s] sir equals the bottom of the test
[87m 24s] yeah so i i was never a fan of the fact
[87m 27s] that you had to your eyes had to check
[87m 29s] that
[87m 30s] where usually your eyes are used to
[87m 32s] checking all of the assertions at the
[87m 34s] bottom
[87m 36s] yeah just
[87m 39s] your three components of a test like a
[87m 41s] wrench act and a third
[87m 43s] switch order so it is a third arrange
[87m 46s] and
[87m 47s] act exactly that makes complete sense
[87m 50s] logically obviously it's going to throw
[87m 51s] an exception so you have to catch that
[87m 53s] earlier rather than later
[87m 55s] and but when you're first new to testing
[87m 57s] you're like well why is everything else
[87m 58s] have to be
[87m 59s] at the end and then everything to do
[88m 1s] with exceptions has to be at the
[88m 2s] beginning
[88m 3s] it's quite nice that that with pest you
[88m 6s] do get to say well actually i want to
[88m 8s] put that at the bottom with all of the
[88m 10s] expectations
[88m 13s] setup actor certainly need adrian
[88m 16s] um okay so if the
[88m 19s] data is a fork
[88m 23s] then we're gonna set the block reason to
[88m 26s] fork but if it's not a fork
[88m 31s] so this is basically only if the user is
[88m 33s] an organization right
[88m 34s] yep um and everything else would be
[88m 38s] block reason
[88m 39s] review if it's a user no if it starts
[88m 43s] with a dot
[88m 45s] if the name starts with the dot yeah
[88m 48s] yeah and then the yeah because
[88m 51s] you're doing and uh safe null check
[88m 54s] there
[88m 56s] very good okay then you can actually try
[88m 59s] to
[89m] create the unique id
[89m 3s] is the id of the data so there's a test
[89m 6s] we could do quite
[89m 7s] easily um which is making sure that you
[89m 10s] can't
[89m 11s] create the same repository twice
[89m 14s] yeah um it's going to insert all that
[89m 18s] data there
[89m 20s] and then if something fails
[89m 24s] which i guess the fail reason as far as
[89m 27s] i can see
[89m 28s] would basically only be because
[89m 31s] some of this data was corrupted
[89m 34s] yeah um if you check what i'm doing
[89m 38s] in this case uh it says only single line
[89m 40s] and it's a
[89m 41s] it's a language attribute so language
[89m 44s] and license are mapped to the
[89m 45s] to an enum instance and
[89m 50s] i don't have a full
[89m 53s] list of all available languages already
[89m 57s] so i already have a lot if you check
[90m 1s] this language enum
[90m 3s] um yeah so 400 lines of languages
[90m 10s] nice but it still could be
[90m 13s] said one is missing and in that case the
[90m 16s] enum will throw a bad method called
[90m 18s] exception
[90m 20s] okay in that case the creation of the
[90m 24s] repository will fail and i will catch it
[90m 27s] and write the tiers's log file missing
[90m 30s] enum values
[90m 32s] and i
[90m 34s] go through this file from time to time
[90m 37s] and
[90m 39s] check which enum values i have to add
[90m 43s] which makes sense so that would that bad
[90m 46s] method call exception will be fired
[90m 48s] if there is a language that doesn't
[90m 49s] exist in your enums
[90m 52s] yeah okay
[90m 58s] but at the end it's more and it returns
[91m] null
[91m 1s] if language doesn't exist
[91m 5s] because the exception is fired but
[91m 7s] catched and the return is now
[91m 13s] so in syria you could even add it to
[91m 17s] to this first test case we already have
[91m 20s] very true so let's
[91m 24s] go to i lost it then but it's right in
[91m 26s] front of me
[91m 28s] set the language do we have valid data
[91m 31s] language set here we don't yet do we
[91m 33s] php for example
[91m 36s] is it or is it usually like uppercased
[91m 39s] yep
[91m 40s] right so let's set the language to
[91m 43s] something crazy
[91m 45s] string random 64.
[91m 50s] certain that there is a language out
[91m 51s] there that is a random 64 character
[91m 56s] string
[91m 58s] that's going to come up one day man and
[92m] you're going to be like there's some
[92m 1s] there's some
[92m 2s] uh some guy with like who always wears
[92m 5s] sunglasses
[92m 6s] creating his own dsl somewhere and
[92m 9s] yeah he names it a 64 character code and
[92m 11s] you're gonna have to
[92m 13s] you're gonna have to add that to the
[92m 14s] venom
[92m 16s] uh awesome
[92m 20s] so that tests for this
[92m 23s] which is nice
[92m 27s] now we can edit a happy path yeah
[92m 31s] who does that i like to just test all
[92m 32s] the nasty things
[92m 35s] uh hit saves
[92m 39s] the repository to
[92m 43s] the owner the owners
[92m 48s] say it links the repository
[93m] okay so in this case we will be saying
[93m 3s] basically repository from github
[93m 7s] this valid data
[93m 12s] and once we've done that we would just
[93m 15s] say
[93m 17s] expect
[93m 21s] repository
[93m 24s] find one
[93m 28s] i don't like this no
[93m 32s] no i don't like ids in in any code
[93m 35s] because if you have for example the
[93m 37s] mysql or postgres driver
[93m 39s] societies won't reset between test runs
[93m 42s] or not between single test cases but
[93m 46s] only
[93m 46s] after every test run um
[93m 50s] so yeah repository first or do a real
[93m 53s] repository the
[93m 54s] name or whatever but
[93m 59s] yeah i mean we could possibly even look
[94m 1s] upon that id right
[94m 4s] yeah yeah the society would work exactly
[94m 7s] we've hard coded that
[94m 8s] id right so what what do we actually
[94m 10s] call that in the repository is it like
[94m 11s] remote id
[94m 13s] no it's just id i see so you actually
[94m 17s] you
[94m 17s] actually set the database id to the
[94m 20s] github id
[94m 22s] auto incremention is disabled and it
[94m 24s] just has
[94m 26s] the github ids yeah
[94m 29s] apparently we're getting no which means
[94m 31s] that obviously at some point
[94m 33s] the logic for that has failed which is
[94m 37s] fine
[94m 39s] yeah it could be that some some valid
[94m 41s] data attributes are missing
[94m 42s] so it's a description key for example
[94m 47s] yeah and let's talk it
[94m 50s] so stark as us count do we even have
[94m 53s] yeah we have a name i'm gonna put the
[94m 55s] name at the top
[94m 57s] much more sense
[95m 1s] a description
[95m 7s] yeah so and a full name
[95m 13s] obviously got ten thousand stars
[95m 18s] [Music]
[95m 20s] the full name is just owner name
[95m 24s] slash uh the repository
[95m 30s] yeah and home page
[95m 34s] can be null but you can also add
[95m 38s] whatever you want
[95m 42s] yeah so matter man
[95m 46s] and now you get a repository instance
[95m 49s] so back to our happy path we would
[95m 52s] expect that
[95m 55s] to be an instance of them uh i'm going
[95m 57s] to do that because it's going to find it
[95m 58s] let's do a find or fail
[96m 1s] because that's going to actually throw
[96m 2s] an exception in our class
[96m 7s] in our test which means we know that
[96m 9s] there is actually a model there it's
[96m 10s] never going to be null
[96m 13s] um and we can just we could just do a
[96m 16s] few tests so we could say
[96m 17s] expect name to be
[96m 20s] um called it php library
[96m 24s] oh it knows the name is comibia slash
[96m 28s] yeah because we assigned the full name
[96m 33s] and description
[96m 39s] it's going to be hello world
[96m 45s] um there's the language
[96m 49s] to be instance of language
[96m 52s] that's a good one yeah and
[97m 2s] yeah and then another one is licensed
[97m 5s] because that's like
[97m 7s] you're going inside two arrays there
[97m 9s] okay
[97m 11s] is that mit lowercase or mit
[97m 13s] applications now license is also
[97m 15s] an enum so license to be instance of
[97m 19s] license
[97m 20s] or otherwise you could also do it to be
[97m 23s] or
[97m 24s] to equal uh license
[97m 28s] color colon uh mit
[97m 31s] yeah let's do that yeah that kind of
[97m 33s] gives us a little bit
[97m 35s] of um i don't know almost
[97m 40s] it's it's an action now you have now you
[97m 43s] have license
[97m 44s] language um it's an extra check
[97m 47s] that we actually that these enums are
[97m 49s] being calculated correctly as well
[97m 51s] without actually writing a test that all
[97m 53s] those enums work
[97m 54s] so could you run it free validation
[97m 59s] perfect so you can use to be also also
[98m 3s] licensed
[98m 5s] yeah you are we have a great
[98m 8s] singleton pattern for the enums
[98m 10s] implemented to be php
[98m 12s] 8.1 future proof yeah because they're
[98m 16s] they're all singletons out the box
[98m 17s] aren't they
[98m 18s] in 8.1 yeah and so we had to to do
[98m 21s] something around it
[98m 23s] and now if you call license mit you will
[98m 26s] always get the exact same object
[98m 28s] reference so you can also
[98m 30s] do something like license mit equals
[98m 32s] equal
[98m 33s] so type save equal so three equals
[98m 36s] uh license mit for example and you will
[98m 38s] get true
[98m 41s] yeah that's that's super useful isn't it
[98m 44s] i think what's nice
[98m 46s] is how close this like i was looking at
[98m 48s] enums literally just yesterday
[98m 50s] for 8.1 and how close the syntax is in
[98m 53s] the
[98m 54s] enum package how close it is in 8.1
[98m 57s] itself
[98m 58s] we did a lot uh of well not a lot but
[99m 2s] several of php 8.1 compatibility
[99m 7s] so polyfills for the other direction
[99m 10s] yeah um so
[99m 13s] you also have access to the try from and
[99m 16s] from methods
[99m 18s] um and
[99m 21s] yeah so you only have to change the enum
[99m 24s] itself
[99m 26s] so we factor from dot block to the case
[99m 29s] notation
[99m 30s] and in your real code
[99m 33s] you only have to change from some
[99m 36s] methods to
[99m 37s] here to the constant like syntax of the
[99m 39s] future enums
[99m 40s] and that's everything you have to change
[99m 42s] because we will keep up the package and
[99m 45s] we also will keep up the level
[99m 47s] enum wrapper for model attribute casting
[99m 51s] and request validation and so on and
[99m 56s] yeah says allows you to already
[99m 59s] use enums now and with very little
[100m 5s] changes you can switch to php 8.1 native
[100m 8s] enums
[100m 10s] that's really nice that's really nice
[100m 12s] yeah that's going to be so helpful
[100m 14s] um yeah so it could be if i if i have
[100m 18s] some time until php 8.1 is released
[100m 21s] it could be that i write a recta plugin
[100m 24s] for example
[100m 25s] uh php rector that will change
[100m 28s] all these method notations of your enums
[100m 32s] to
[100m 32s] the constant lagrange so you can call
[100m 35s] the recta
[100m 36s] and on your whole code base and
[100m 38s] independent if you
[100m 39s] have one enum in one place or a million
[100m 42s] enoms in 10 million places
[100m 45s] it will be done in like a minute
[100m 48s] yeah that's yeah i mean
[100m 52s] yeah i i've not used rector but
[100m 54s] certainly it seems possible but it's
[100m 56s] close enough that it wouldn't actually
[100m 57s] be a massive job i don't imagine to
[100m 59s] to write the script for that yeah
[101m 8s] we've got a little bit of uh
[101m 12s] resource path of course
[101m 16s] yeah so i have
[101m 20s] sadly taken that a little bit too far
[101m 24s] uh i can't actually think of
[101m 27s] weight until 8.1
[101m 30s] i can't think of a way to do that
[101m 34s] in higher order because it's failing
[101m 37s] basically because of the
[101m 41s] resource path is here right yep
[101m 49s] i was okay let's just go back a second
[101m 51s] let's let's at least
[101m 53s] verify that i am correct in saying
[101m 57s] it's not possible
[102m] i'm going to comment those two lines out
[102m 4s] yeah okay so something in the enums
[102m 9s] is is obviously performing a
[102m 15s] some form of like look up into the
[102m 17s] container into my rolls container oh yes
[102m 20s] the language one
[102m 22s] so yeah if that was coming if you if you
[102m 24s] open the language one yeah
[102m 27s] you have this values and
[102m 31s] here you have to collect json decode
[102m 33s] file get resource path
[102m 36s] yeah which is fair news um i think
[102m 40s] in that case we just move this back into
[102m 42s] a
[102m 43s] standard expectation then
[102m 47s] uh awesome
[102m 50s] okay so there's a happy path we have a
[102m 53s] happy path that actually checks that
[102m 54s] repository is created
[102m 56s] i guess the other thing we could do is
[102m 59s] uh
[102m 59s] let's we could make this we could
[103m 4s] continue the chain but because i'm going
[103m 5s] to be testing two different things i'm
[103m 7s] going to bring this out
[103m 9s] into its own variable
[103m 13s] and i will run a bunch of checks on the
[103m 15s] repository but then i will say
[103m 17s] repository owner and i'll run
[103m 25s] i'll make video and i have its own
[103m 27s] expectations
[103m 29s] because this owner doesn't exist at the
[103m 31s] start of this test
[103m 32s] so it's actually creating the gummy bear
[103m 35s] user at the same time isn't it
[103m 37s] yeah it's kind of like a nice bit of
[103m 40s] validation at the same time
[103m 41s] that all works really nice
[103m 45s] awesome i'll inline that since we're
[103m 47s] doing a single statement so
[103m 50s] you can't have two levels of higher
[103m 52s] order before you do
[103m 53s] expectations
[103m 56s] uh sorry so you can't
[103m 59s] you so you have the repository in the
[104m 2s] first expectation
[104m 4s] and i sincerely you could chain just
[104m 7s] owner
[104m 8s] name to be gamibia but i don't know if
[104m 10s] these
[104m 11s] two levels uh are possible yeah they are
[104m 14s] so that i
[104m 15s] specifically like went in and refactored
[104m 18s] the higher order expectation api to
[104m 20s] support
[104m 21s] nested so that does work i'll show you
[104m 24s] okay
[104m 25s] and you can also do before uh
[104m 29s] before the owner name check you can do
[104m 31s] an owner to be
[104m 33s] a user i think we have named it
[104m 36s] so one we have uh no not instance off
[104m 41s] the cusp and customer expectation we
[104m 43s] have added yeah yeah
[104m 44s] look at that by the way i've seen this
[104m 45s] in the in the php
[104m 48s] in the pest plug-in the latest version
[104m 51s] you get
[104m 51s] type you get completion for your expect
[104m 53s] extensions
[104m 55s] perfect which is real nice like i don't
[104m 58s] know how oliver did that
[104m 59s] some magic going on there but yeah you
[105m 1s] you get a type completion for all of
[105m 3s] your
[105m 3s] expect extensions
[105m 8s] awesome yeah that still passes so i i
[105m 11s] guess what i was saying is like
[105m 13s] if i was going to do a lot of expecting
[105m 15s] on the user model
[105m 18s] i would i would not do this in a single
[105m 21s] expectation query and i would break it
[105m 24s] into two because i think it
[105m 26s] it kind of starts to become a little bit
[105m 30s] unreadable you don't know where one one
[105m 32s] ends and the other begins
[105m 35s] um but if we're only checking that it's
[105m 36s] a user and the name is gummy beard then
[105m 40s] yep okay it has been has gone so fast
[105m 45s] but it has been
[105m 46s] one and three quarter hours on the
[105m 48s] stream there
[105m 49s] um that's been awesome though that's
[105m 52s] been fun to kind of
[105m 54s] really go into something with a lot of
[105m 56s] logic going on well
[105m 58s] a fair bit of logic going on and be able
[106m] to use some stuff like
[106m 2s] uh data sets to loop over and change
[106m 4s] data slightly
[106m 5s] and tweak it for every possible scenario
[106m 9s] and yeah man that's been fun
[106m 12s] thanks everyone for joining thanks
[106m 13s] everyone who's uh
[106m 16s] been kind enough to to stay and listen
[106m 19s] for that long as well i
[106m 20s] didn't mean for the stream no one as
[106m 21s] long as it has so i really appreciate
[106m 23s] that
[106m 24s] if you've got any questions pest related
[106m 26s] questions
[106m 27s] then be sure to send them to tom on
[106m 30s] twitter
[106m 31s] he would love to
[106m 35s] send something you know only call me if
[106m 37s] you have if you have append or leg
[106m 39s] collisions
[106m 43s] um it's pretty cool we're relaxed in the
[106m 47s] speed at which we're moving here but to
[106m 50s] be able to sit back and think of
[106m 52s] different ways to write tests for
[106m 55s] something that is already in production
[106m 56s] is always interesting
[106m 59s] making sure that you kind of catch every
[107m 1s] scenario but still write in a way that
[107m 4s] is open for extension further down the
[107m 6s] line and isn't going to completely break
[107m 8s] down when you go and make a change
[107m 10s] um so yeah that's always fun um
[107m 14s] yeah all right so uh i assume
[107m 17s] i've not actually asked you yet tom but
[107m 18s] i assume we're good for next week
[107m 21s] and we can carry this this thing on
[107m 24s] probably
[107m 25s] either look a little bit more into this
[107m 27s] repository class and add some more tests
[107m 30s] that are based on repository or will
[107m 31s] come back up the chain
[107m 32s] into that job and possibly move along
[107m 36s] the other jobs
[107m 37s] but i i will decide that before next
[107m 39s] week i'll have a look through and see
[107m 40s] which one
[107m 41s] looks more interesting for the stream
[107m 44s] um as always i will push this up right
[107m 48s] now
[107m 49s] and this is going to be available
[107m 52s] on in the pr on the open door dot me
[107m 56s] github
[107m 58s] so if you want to go see the code if you
[108m] want to edit the code
[108m 2s] and suggest a better way of doing things
[108m 4s] then go check it out
[108m 5s] it is there and available for you to do
[108m 7s] as you please
[108m 9s] but with all that said and done i'm
[108m 11s] going to
[108m 12s] stop sharing that and i'm going to say
[108m 15s] goodbye
[108m 16s] and thank you very much tom for joining
[108m 18s] us again
[108m 20s] thank you and for your time and
[108m 23s] contributions
[108m 25s] and see you again next week
[108m 28s] i hope so yeah yeah
[108m 37s] as you can see one last exciting thing
[108m 40s] which hopefully might be there for next
[108m 41s] week although
[108m 42s] no promises is uh
[108m 46s] you mentioned about the repeat
[108m 48s] functionality in in pest
[108m 51s] so this came up on twitter but just to
[108m 54s] kind of show people what what we're
[108m 57s] talking about
[108m 58s] imagine being able to say like imagine
[109m 1s] you're working with a data
[109m 2s] like a piece of functionality that does
[109m 4s] something quite complex and
[109m 6s] every so often your ci fails right your
[109m 9s] continuous integration fails
[109m 11s] and breaks down and there's a test that
[109m 13s] fails on you and you have no idea why
[109m 16s] it's failed
[109m 16s] and then you run the ci again and it
[109m 18s] passes
[109m 20s] i don't know whether you've ever had
[109m 21s] that scenario but i have had that before
[109m 24s] so one thing that we're i'm working on
[109m 26s] currently for pest which would be
[109m 28s] awesome if we can get it in
[109m 29s] it's just a repeat function and you can
[109m 31s] pass any number into repeat
[109m 34s] and the test will be repeated as its own
[109m 36s] test so it's not
[109m 38s] like each test will still get its own
[109m 40s] instance it will get its own
[109m 42s] database set up so everything will be
[109m 44s] fresh each time
[109m 45s] and it will just run that test x amount
[109m 47s] of times and you're going to get
[109m 49s] like console output matches as well um
[109m 52s] so that's what's being worked on and
[109m 54s] that will come in super useful because i
[109m 56s] i have had that situation before where
[109m 59s] you've
[109m 59s] forgot to update a fixture somewhere or
[110m 1s] there's something that you've not
[110m 3s] factored in
[110m 4s] and yeah once every blue blue moon the
[110m 6s] test breaks
[110m 8s] um yeah or your factory for example
[110m 11s] doesn't
[110m 12s] hundred percent match
[110m 15s] all of your business logic like for
[110m 17s] example
[110m 18s] you generate a product price between
[110m 20s] zero and one hundred
[110m 22s] and in one percent of the cases it
[110m 24s] exactly hits the zero
[110m 27s] and you have some special conditions
[110m 29s] like products have to cause something or
[110m 32s] if they don't cost anything they aren't
[110m 34s] audible or whatever things
[110m 36s] and now your test breaks and
[110m 39s] yeah this will happen once
[110m 43s] per year for example but you would drive
[110m 45s] crazy
[110m 46s] yeah yeah yeah absolutely and
[110m 50s] it guaranteed it always happens when
[110m 51s] you've like got a deadline
[110m 53s] next time yeah all right and it's
[110m 56s] capable of only hit the ci button run
[110m 58s] again okay green
[111m] let's merge it yeah
[111m 3s] but it's like one of those things now
[111m 5s] where instead of having to
[111m 7s] like run the test and then wait and run
[111m 10s] it again
[111m 10s] and then run it again and keep doing
[111m 11s] that you're just going to be able to
[111m 13s] drop a repeat in there drop however many
[111m 16s] times you want that test to run
[111m 18s] and you can either remove it after
[111m 19s] you're finished or leave it in there for
[111m 21s] some assurance
[111m 22s] so yeah that that should be coming soon
[111m 24s] i've just been
[111m 26s] um i think yeah it was oliver who
[111m 29s] suggested that the
[111m 30s] the actual output moved from the
[111m 33s] description name which is where
[111m 34s] i originally put it into the console
[111m 37s] output itself
[111m 38s] so it's basically handled by the test
[111m 41s] printer rather than
[111m 42s] the test suite so that that requires
[111m 46s] going
[111m 46s] and creating a pr for um
[111m 50s] for collision which is like the package
[111m 52s] that pest uses
[111m 54s] to output it's pretty
[111m 57s] text so yeah once that's done then the
[111m 59s] repeat functionality will go in i think
[112m] that's really nice
[112m 2s] um so yeah
[112m 5s] i've held myself to that now there we go
[112m 10s] awesome well thanks everyone for joining
[112m 11s] us and we'll see you next week
[112m 14s] yeah see you bye