Closure Compilation builder is broken |
||||||
Issue descriptionThis is causing failures on https://build.chromium.org/p/tryserver.chromium.linux/builders/closure_compilation /b/build/scripts/slave/.recipe_deps/depot_tools/ninja -w dupbuild=err -C /b/build/slave/linux/build/src/out/Release ninja: Entering directory `/b/build/slave/linux/build/src/out/Release' ninja: fatal: chdir to '/b/build/slave/linux/build/src/out/Release' - No such file or directory [Errno 2] No such file or directory: '/b/build/slave/linux/build/src/out/Release/.ninja_log' step returned non-zero exit code: 1 @@@STEP_FAILURE@@@ Looks like it's still expecting the GYP out dir? Steven, do you know what's going on?
,
Sep 14 2016
Closure is still using GYP; migrating Closure to GN is a separate project and is not completely straightforward. It's not clear to me what that output snippet has to do with closure though.
,
Sep 14 2016
It seems that it's trying to access the directory src/out/Release which I think we don't use anymore since we switched to GN, and it won't be generated automatically as a result of sync or runhooks since we disabled that, correct? I'm not sure if that builder assumes the generation of the GYP files automatically somehow.
,
Sep 14 2016
With GN the directory can be any name. I don't know that we changed it on the builders, i.e. they may very well just use out/Release.
,
Sep 14 2016
It looks like the last successful release used out/Release: https://build.chromium.org/p/tryserver.chromium.linux/builders/closure_compilation/builds/2669/steps/compile/logs/stdio
,
Sep 14 2016
It looks like a bunch of build slaves on this trybot have never done a successful closure_compilation (or anything at all), I wonder if they are misconfigured?
,
Sep 14 2016
The generate_gyp_files stage does not appear to have a lot of useful output (like what directory it is writing to), we might want to start by improving that output. The recipe is here: https://cs.chromium.org/chromium/build/scripts/slave/recipes/closure_compilation.py?q=generate_gyp_files&sq=package:chromium&dr=C&l=43
,
Sep 15 2016
ah, yeah, this is probably a runhooks-related thing see also: bug 646969
,
Sep 15 2016
this is breaking people's ability to commit
,
Sep 15 2016
How is that breaking people's ability to commit, unless they include closure_compilation in CQ_INCLUDE_TRYBOTS?
,
Sep 15 2016
also, I think you listed the wrong bug in #8 :).
,
Sep 15 2016
I think when we deleted common.gypi you lost the default configuration info and so things are going into out/Default instead of out/Release, but of course ninja doesn't know that. For now I'll re-add common.gypi and put the line back into gyp_chromium that you need, and hopefully that'll fix it until I have time for a better fix.
,
Sep 15 2016
dpranke@, right bug 632206 is what i meant
,
Sep 15 2016
,
Sep 15 2016
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/0109afc37c2d185d9825e4a93809e4acab435a23 commit 0109afc37c2d185d9825e4a93809e4acab435a23 Author: dbeam <dbeam@chromium.org> Date: Thu Sep 15 04:25:39 2016 Re-add common.gypi to help fix closure_compilation. This is a temporary fix until Dirk can change the closure_compilation uses of gyp_chromium correctly. Stolen from Dirk Pranke <dpranke@chromium.org> and stripped of 'includes': https://codereview.chromium.org/2331423005/ TBR=brettw@chromium.org, dpranke@chromium.org CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:closure_compilation BUG= 646969 Review-Url: https://codereview.chromium.org/2345503005 Cr-Commit-Position: refs/heads/master@{#418779} [add] https://crrev.com/0109afc37c2d185d9825e4a93809e4acab435a23/build/common.gypi [modify] https://crrev.com/0109afc37c2d185d9825e4a93809e4acab435a23/build/gyp_chromium.py
,
Sep 16 2016
This particular bug is fixed. The underlying issue (needing to move the Closure builders off of GYP) is not, but that's tracked in bug 632206 .
,
Sep 16 2016
If so, let's rename this issue appropriately and mark it as fixed. |
||||||
►
Sign in to add a comment |
||||||
Comment 1 Deleted