reosarevok: maybe I've forgotten what the issue is exactly, but why can't we do whatever it previously did so that's it's equally-broken as before instead of more-broken?
2021-09-28 27104, 2021
reosarevok
Earlier we didn't link to the reltype at all
2021-09-28 27116, 2021
reosarevok
It seems useful to do so for the 90%+ that are not broken
2021-09-28 27141, 2021
reosarevok off to the shop, but biab and will look at open PRs in the milestone, change the collections stats one, then release beta tonight again for freeze
2021-09-28 27101, 2021
yvanzo
which milestone?
2021-09-28 27118, 2021
yvanzo
Oh ok, beta has not been updated. So releasing beta at night so no one will notice if it’s broken and in which way... looks like a good plan ;)
2021-09-28 27115, 2021
yvanzo
bitmap: It seems the special commands we have for Selenium could be implemented as plugins now, using "execute script", right?
2021-09-28 27128, 2021
yvanzo
Does it seem worth doing that to allow using Selenium IDE again?
2021-09-28 27110, 2021
yvanzo
Or evaluate moving to something else like Nightwatch.js?
2021-09-28 27159, 2021
bitmap
I'm not sure how you'd implement things like handleAlert or mouseOver in the browser (events in general might be iffy if the browser doesn't see them as "native")
2021-09-28 27111, 2021
bitmap
but some of them should be possible
2021-09-28 27128, 2021
bitmap
I've never heard of nightwatch so it's not clear to me how that helps
2021-09-28 27114, 2021
bitmap
can it run selenium IDE files directly?
2021-09-28 27137, 2021
param
o/
2021-09-28 27142, 2021
param
how goes it, friends
2021-09-28 27151, 2021
ruaok
hey, how are you?
2021-09-28 27117, 2021
param
good good, busy few months at work. :/
2021-09-28 27124, 2021
param
what'd i miss :D
2021-09-28 27129, 2021
ruaok
a lot. :)
2021-09-28 27138, 2021
param
as expected haha
2021-09-28 27156, 2021
yvanzo
hi param!
2021-09-28 27117, 2021
param
hey yvanzo!
2021-09-28 27128, 2021
ruaok
LB has had a lot of new features. brainzplayer is usable, you can play your love/hate, similar users now works well and finds users. we're really making good progress.
2021-09-28 27140, 2021
yvanzo
bitmap: IIUC, it is higher level than Selenium, might be able to replace some of our customizations (?).
2021-09-28 27100, 2021
param
i saw how good similar users has gotten!
2021-09-28 27116, 2021
param
it found a few friends of mine who used to listen to the same music with me in college lol
2021-09-28 27129, 2021
param
very impressive.
2021-09-28 27154, 2021
ruaok
it was instumental to have Kirsten as such a close match -- if we weren't near the best match, clearly something was off.
did you want to update the query in count.collection.public to use <<~SQL too?
2021-09-28 27126, 2021
reosarevok
I guess that'd be consistent
2021-09-28 27105, 2021
reosarevok
Oh, there's also perlcritic issues, heh, will fix
2021-09-28 27140, 2021
yvanzo
bitmap: handleAlert is used in 5 tests, mouseOver in 1. Might it be the majority of tests can be made without customizations? Would it be useful to have a conversion script side->json5 for those? (the opposite of https://gist.github.com/y-young/b753b2d2ea1bab21d… )
2021-09-28 27125, 2021
bitmap
yeah, I think a conversion script would be useful
2021-09-28 27109, 2021
bitmap
iirc the IDE would tend to use weird or unstable xpath selectors if you clicked on something, so I would usually hand-edit some things too
2021-09-28 27118, 2021
lucifer
monkey: hi! thoughts on how to make UserStats component more modular? i'd like to reuse it for sitewide stats.
2021-09-28 27159, 2021
reosarevok
bitmap: I guess there's no way to just separate the timeout for tests and the timeout for DB clearing?
2021-09-28 27144, 2021
bitmap
i'd prefer that but it's a bit more complicated
2021-09-28 27153, 2021
bitmap
let me see again
2021-09-28 27149, 2021
bitmap
also, I increased the timeout on jenkins from 15 minutes to 20 minutes
most of the page can be the same i think. yes UserEntityChart is a part of it.
2021-09-28 27101, 2021
riksucks
hey param!
2021-09-28 27155, 2021
tandy[m]
<param> "i saw how good similar users has..." <- it managed to spot me and my partner lol
2021-09-28 27156, 2021
tandy[m]
pretty good
2021-09-28 27127, 2021
tandy[m]
only thing that would be neat is to show user similarity when viewing another users page
2021-09-28 27133, 2021
reosarevok
bitmap: is it mostly because of the db clearing? Or does it just time out because of actual test stuff taking a long time?
2021-09-28 27150, 2021
yvanzo
bitmap, reosarevok: IIRC we used to want adding selenium tests to help with the conversion to react mainly. Now that TT is mostly gone anyway, it seems many tests could be made using enzyme ^ instead? Using Selenium just when needed to.
2021-09-28 27129, 2021
tandy[m]
i was wondering, is there a timeline for this issue?
yvanzo: they don't really test the same thing though...enzyme helps you test a component in isolation, but doesn't tell you if the actual page works. I think there's a lot of value in the selenium tests
2021-09-28 27151, 2021
lucifer
its mostly a rewrite using a authlib, so that we can get rid of our own impl.
2021-09-28 27138, 2021
yvanzo
bitmap: That's the point, we don't have any test for component in isolation. Thinking about it as complementary, not a full replacement.
2021-09-28 27139, 2021
lucifer
regarding BP embed, monkey might be able to help you. there were explorations into it earlier in the year but we hit some issues. (do not recall exactly what)
2021-09-28 27152, 2021
lucifer
the similarity to other user thing sounds good. we should probably add it.
2021-09-28 27135, 2021
lucifer
ruaok: thoughts? displaying how similar another user is to you when you visit their page?
2021-09-28 27131, 2021
lucifer
i had thought of displaying that in the search user name results as well but figured our way to store similarity might not be performant for those queries.
2021-09-28 27135, 2021
bitmap
yvanzo: sure, if by "it seems many tests could be made using enzyme ^ instead" you meant adding new tests instead of converting existing ones
2021-09-28 27128, 2021
bitmap
enzyme seems useful for smaller components used on tons of pages, like EntityLink
2021-09-28 27141, 2021
yvanzo
bitmap: yes, for example merge redirect cannot be tested using enzyme.
2021-09-28 27128, 2021
yyoung[m]
IIRC some commands in IDE are not supported in our tests
2021-09-28 27132, 2021
bitmap
anything that submits data, really :)
2021-09-28 27111, 2021
reosarevok
bitmap: huh, why do we have sir for our selenium tests?
2021-09-28 27127, 2021
yyoung[m]
And yes the selectors generated by recording the actions are not reliable
2021-09-28 27155, 2021
bitmap
reosarevok: because it's testing the entire UI and makes use of indexed search in some places
2021-09-28 27106, 2021
yvanzo
(to fill the search indexes)
2021-09-28 27121, 2021
reosarevok
I would have expected us to just mock that since it's not really MB related at ll
2021-09-28 27121, 2021
reosarevok
But ok
2021-09-28 27128, 2021
reosarevok
well, MB*S* related :)
2021-09-28 27109, 2021
yvanzo
From mock tests in sir, I guess you don't want that for MBS ;)
2021-09-28 27154, 2021
reosarevok
Well, I don't want to mock MBS workings in MBS
2021-09-28 27113, 2021
reosarevok
But it seems fine to test sir inside sir, and just mock what it should respond with for MBS
2021-09-28 27136, 2021
bitmap
it's an integration test
2021-09-28 27119, 2021
bitmap
(testing whether all the components work when combined together)
2021-09-28 27147, 2021
reosarevok
Well, if it doesn't make things a lot slower, I guess it doesn't matter - just wondering whether this is something we could break by changing MBS (because if it is slow and runs for every testing run, that's annoying)
2021-09-28 27158, 2021
reosarevok
(if it's something that doesn't depend on our changes, I mean)
2021-09-28 27125, 2021
reosarevok
(I'm not trying to be annoying, maybe I'm just slow :) )
2021-09-28 27109, 2021
yvanzo
Mocks might be useful, but integration tests are needed too.
2021-09-28 27127, 2021
bitmap
well the point is just that mocking sir doesn't test the same thing
2021-09-28 27126, 2021
bitmap
so if it makes the tests slower I'd rather focus on making them faster than removing the tests :)
2021-09-28 27119, 2021
reosarevok
Sure, I'm just trying to understand whether this sort of thing should be tested every time we do an MBS PR, or if it should be tested as a separate thing because it's about how sir answers to a specific MBS call. But I guess the point is that it will tell us if something has changed in sir that breaks our stuff?
2021-09-28 27125, 2021
reosarevok
Anyway, if it's useful it's useful :)
2021-09-28 27130, 2021
monkey
lucifer: please remind me tomorrow about the charts component, can't think right now :)
2021-09-28 27136, 2021
lucifer
sure
2021-09-28 27108, 2021
yvanzo
reosarevok: It can break in so many ways, depending how mbs/sir/mbsssss/mbdata/mb-rng/mb-solr are combined, it is worth having this integration test. :)
yvanzo: oh, target, yeah. I guess for that one it should be target URL?
2021-09-28 27126, 2021
reosarevok
(re: the other PR)
2021-09-28 27147, 2021
bitmap
it looks fine to me other than the target issue anyway
2021-09-28 27135, 2021
yvanzo
reosarevok: I would prefer not systematically moving the leftovers of a milestone to the next one but rather reevaluate PRs for each new milestone; What do you think?
2021-09-28 27142, 2021
reosarevok
I'm still not 100% on what each target type is for (when to use RELATIONSHIP for example, and NONE)
2021-09-28 27153, 2021
reosarevok
Sure, I guess, if you feel some of the bits just aren't ready for next week
2021-09-28 27100, 2021
reosarevok
re: reevaluating :)
2021-09-28 27119, 2021
piwu has quit
2021-09-28 27138, 2021
yvanzo
reosarevok: about the error target, I’m not sure if your code allows for a specific target actually, might need to be split otherwise.
2021-09-28 27156, 2021
yvanzo
and yes the PR looks fine otherwise
2021-09-28 27114, 2021
reosarevok
So, ENTITY seems to be "this is fine, but not for this entity"
2021-09-28 27124, 2021
reosarevok
URL is I guess "this would be fine but the URL is borked"?
2021-09-28 27130, 2021
reosarevok
What about RELATIONSHIP and NONE?
2021-09-28 27146, 2021
yvanzo
+1
2021-09-28 27153, 2021
yvanzo
It seems it should be RELATIONSHIP if the type id doesn’t match (switch cases), probably URL otherwise?
2021-09-28 27101, 2021
piwu joined the channel
2021-09-28 27104, 2021
reosarevok
Not ENTITY? Since the URL is probably valid if you were just adding it to a recording or release?
