samth
2019-11-26 15:00:57

What’s the best way to use xvfb on GHA?


samth
2019-11-26 15:03:52

Also, my current attempt to use GHA for typed-racket takes about 3x longer to set up than on travis, which is problematic


samth
2019-11-26 15:04:12

I’m doing it slightly differently, but I am surprised by the difference


deactivateduser60718
2019-11-26 15:15:40

@deactivateduser60718 has left the channel


samth
2019-11-26 15:18:20

@popa.bogdanp I think part of the difference is your setup-racket action installs globally whereas previously I was installing “in-place”


pocmatos
2019-11-26 15:19:54

When running pkgs/racket-test-core/tests/racket/test.rkt, is there a way to disable specific sections?


samth
2019-11-26 15:21:22

no, the best way to disable specific sections is to run all the other sections one-by-one


samth
2019-11-26 15:32:21

ah, things ran much faster when I switched to the way I did it on Travis


pocmatos
2019-11-26 16:09:55

But on Travis is, I think, how I am doing it for cgc at the moment.


popa.bogdanp
2019-11-26 17:23:44

I assume you’re referring to this change:

https://github.com/racket/typed-racket/commit/45a7fdba0ae6834fdc82e767bb8800b9d670058d

Is it the combination of -i , --no-setup and --skip-installed that gets you the speed gains? Why is it necessary to create a custom catalog?


samth
2019-11-26 17:24:50

Yes, that change. I’m not totally sure, but I think it’s mostly not setting up all the packages that depend on TR. That’s accomplished with --no-setup and then running raco setup after.


samth
2019-11-26 17:25:10

And if there’s a better way than that custom catalog hack I’m happy to hear it.


samth
2019-11-26 17:25:27

Also, that might be something worth integrating into your Action.


samth
2019-11-26 17:26:28

@pocmatos it looks like the ‘upload-artifact’ action will zip for you if you give it a directory


popa.bogdanp
2019-11-26 17:27:35

Yeah, it would be nice if the action could do most of this for you. I’ll look into it this weekend


samth
2019-11-26 17:28:01

It might be that just the raco pkg update could work without the catalog.


samth
2019-11-26 17:28:08

I should try that.


samth
2019-11-26 18:49:47

@pocmatos can you file a bug listing the tests that fail under cgc? Also, I expect testing on 3m to usually be the better/more important/more likely to work version.


pocmatos
2019-11-26 19:25:05

Not sure you mean about the upload-artifact but it’s already used to build 3m and cs from cgc.


pocmatos
2019-11-26 19:25:29

Yes - thats the idea, to file a bug for all flaky tests with a reference to the reverted commit.


pocmatos
2019-11-26 19:25:44

I am in the middle of getting my kids in bed so I will either do it later on or tomorrow morning.


pocmatos
2019-11-26 19:27:22

Also send the meeting report to Jack to review and will post it tomorrow if he has nothing to add. Maybe all of us (at the moment it seems its me, you, @deactivateduser60718 and @notjack) should try to agree on a time to meet to discuss further steps. The very near-future steps I agree with Jack will be in the report I wrote.


pocmatos
2019-11-26 19:27:43

Oh, it seems Sage left the channel.


pocmatos
2019-11-26 19:30:33

Also relevant is how to deal with the spread of source code over several repos and create a unified CI story. For example, should we trigger Typed Racket tests when Racket code changes? Should we trigger Racket tests (in CS mode) when Chez code changes? I assume we need to repo-local triggers to do this.


pocmatos
2019-11-26 19:30:42

go to go.


deactivateduser60718
2019-11-26 20:08:21

@deactivateduser60718 has joined the channel


notjack
2019-11-26 20:08:40

I have lots of thoughts on cross-repo CI but they can wait until after the immediate next steps


deactivateduser60718
2019-11-26 20:09:53

Oh, it seems Sage left the channel. I stepped out since the timing didn’t seem right for what I thought I was offering. I just like keeping my channels lean, and I am still available.


deactivateduser60718
2019-11-26 20:12:37

Current short-term availability (Eastern Time) is all day tomorrow the 27th and Friday morning the 29th. Not available on the weekend.


pocmatos
2019-11-26 20:56:11

@notjack if you get 5 mins, can you please take a look at the briefing I sent you of the meeting we had? Thanks.


notjack
2019-11-26 20:56:53

@pocmatos Can do, but not until later today.


pocmatos
2019-11-26 21:08:34

That’s fine. Thanks


pocmatos
2019-11-26 21:19:09

@samth Can’t seem to run a single test section - do you know which command line I need to use?


samth
2019-11-26 21:19:48

[samth@huor:~/.../racket-test-core/tests/racket (master) plt] racket -f quiet.rktl vector.rktl Section(vector) SUMMARY ---------------------------- Performed 1602 expression tests (1406 value expressions, 196 exn expressions) and 392 exception field tests. Passed all tests.


pocmatos
2019-11-26 21:20:05

Ah! -f. Was using -l.


pocmatos
2019-11-26 21:20:13

need to check what -f is all about…


samth
2019-11-26 21:20:16

yeah, those files need to be run with load


pocmatos
2019-11-26 21:20:37

ah… interesting, always though -l was for load.


pocmatos
2019-11-26 21:20:52

turns out -l is for lib and -f is for load. TIL


pocmatos
2019-11-27 07:42:33

I was just trying out tests on section expand and got:


pocmatos
2019-11-27 07:42:41

Section(expand) ERROR: open-input-file: cannot open input file path: /home/runner/work/racket/racket/pkgs/racket-test-core/tests/racket/mz-tests.rktl system error: No such file or directory; errno=2


pocmatos
2019-11-27 07:42:49

mz-tests.rktl doesn’t seem to exist.


pocmatos
2019-11-27 07:43:01

Am I missing something or is this the case for a new Issue.


pocmatos
2019-11-27 07:43:04

?