Is the clang in /release+asserts/ really +asserts ?

[Thomas Sepez]

So anyways, the last time I built the chrome plugin and ran it against a locally built clang, I encountered a bug in llvm where it was tripping an assert[1].

Using the pre-built compiler that eventually rolled to include the plugin changes, I can now build cleanly without tripping that assert.

A quick perusal didn't show any changes to clang which might have fixed the issue, and my pull request is still stalled[2].

I was wondering if, by chance, the asserts were not really present for our production builds, and whether it was OK to assume that they would remain so (seems doubtful), but  that would unblock some work.

Otherwise, if assets are expected in the prebuilt compiler, I'd need to dig further into what happened.

Thanks!

1. https://github.com/llvm/llvm-project/issues/119786

2. https://github.com/llvm/llvm-project/pull/124022

[Nico Weber]

No, it's built without asserts since quite a while ago. But the name of that directory is in a few places and a bit hard to change. There's a bug for that with more details somewhere.

Sorry!

--
You received this message because you are subscribed to the Google Groups "cxx" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cxx+uns...@chromium.org.
To view this discussion visit https://groups.google.com/a/chromium.org/d/msgid/cxx/CAPYU9wRxBGaTO1Wsr-WhFczpjhxSWUAUcH3v1H1crmuxen8NNA%40mail.gmail.com.

[Hans Wennborg]

Just to add: we do toolchain testing with asserts enabled, so tripping an assert would break that process.

To view this discussion visit https://groups.google.com/a/chromium.org/d/msgid/cxx/CADZ1XiZ3DzBzSxHr_sxEoVVGmLLtB_D7WtFfZMWo80Ad_myGDQ%40mail.gmail.com.