qt5.{qtwebengine, qtwebkit}: fix -Wno-class-memaccess warning for clang #70544
Conversation
|
Links to broken builds: https://hydra.nixos.org/build/102530965 |
veprbl
added
the
9.needs: port to stable
| @@ -54,9 +54,9 @@ qtModule { | |||
| [ | |||
| # with gcc7 this warning blows the log over Hydra's limit | |||
| "-Wno-expansion-to-defined" | |||
There was a problem hiding this comment.
But this one gets ignored?
There was a problem hiding this comment.
No, this one is actually supported in clang: https://clang.llvm.org/docs/DiagnosticsReference.html#wexpansion-to-defined
There was a problem hiding this comment.
To clarify my phrasing/use of "ignore" - with GCC it's the default to ignore non-existent warning options. With clang it's afaik not. There is -Wno-unknown-warning-option for that, which we don't use as far as I can tell.
|
@GrahamcOfBorg build qt5.qtwebengine |
|
hmm, I guessing will probably timeout on darwin and ofborg in general. A darwin tester is probably needed (or builder access). |
|
@worldofpeace: I briefly tried it out running the beginning of the build on NixOS by overriding |
|
Ahh @d-goldin, I guess it's sufficient to switch it out for clang. Will try that 👍 |
|
Hmm, I'm not sure how I can do that with these qt expressions... |
|
@worldofpeace: I tried it with |
|
Uhm, but you know what - I looked at the log again, and I got misled by the warning about the wrong flag. This does not actually seem to cause issues with the build itself (at least log length due to warning does not seem problematic either at this stage) so I guess it's a "nice to make it cleaner" but did not actually break the builds. Really sorry for the hassle, my bad :( |
d-goldin
changed the title
With a previous fix for log size issues due to GCC 8 a gcc specific `-W` flag was added that clang does not know, so it spams the logs.
d-goldin
force-pushed
the
fix_qtweb_warning_clang
branch
from
fa3375a
to
3bfe087
Compare
|
backported in 9bbad4c |
Motivation for this change
With a previous fix for log size issues due to GCC 8.a gcc specific
-Wflag was added that clang doesnot gracefully ignore
See comment #70544 (comment)
This does not break the build, it just logs unneeded warnings and the breakage is somewhere else.
This was done in #68434 and #68869
We forgot to test this with clang and/or on darwin.
cc @worldofpeace
Things done
sandboxinnix.confon non-NixOS)nix-shell -p nix-review --run "nix-review wip"./result/bin/)nix path-info -Sbefore and after)Notify maintainers
cc @