LNT bot failures


I'm scratching my head to understand what's wrong with this bot:


There were no recent commits to lnt or the test-suite and I've cleaned
the lnt.venv on the bot, but it's still failing. I also haven't
updated the packages or anything on the bot.

Can any LNT specialist have a look and see if I'm missing anything obvious?


To me, that looks like LNT is having problems parsing the version info out of the compiler binary.

LNT executes:

   cc_version = capture([cc, '-v', '-E'] + cc_flags +
                         ['-x', 'c', '/dev/null', '-###'],

Then parses the result

Perhaps that output has changed in an unexpected way?

Looks like you fixed it. The regex was not finding the compiler version because the space after -### was missing.

I have a fix here that I will commit which makes not finding the compiler version a hard error. Since what we were doing was printing an error, then crashing, a sad exit seems like a better idea.

See cfe-commits thread:

"Re: r214924 - Make crash diagnostics on Windows the tiniest bit more useful"

It seems a removed space broke our version parsing.


I just deleted the source dir, and the test-suite dirs too. Not sure
how that would have fixed the problem.


Reid reverted the unintentional whitespace change. That's what made it work again.


Ah, coincidence. :slight_smile: