Nico Weber
Nico Weber
Let's wait for #22325 and then take the parts that convert actual strings to String :)
@alimpfard I guess the thing to do here is to rename DeprecatedFlyString to FlyBytes or some such? (As-is, the rhs has very different performance characteristics than the lhs, and this...
Importing upstream as one commit seems alright, but maybe changes from upstream should be in separate commits?
From the upstream issue, the correct demangling is ``` Undecoration of :- "??$emplace_back@AEBM@?$vector@MV?$allocator@M@std@@@std@@QEAA?A_TAEBM@Z" is :- "public: decltype(auto) __cdecl std::vector::emplace_back(float const & __ptr64) __ptr64" ``` apparently.
How did you end up with these symbols? Is that from building a binary with cl.exe, or with clang-cl.exe? I'm wondering if clang-cl also mangles it incorrectly, in addition to...
~~For this https://godbolt.org/z/6env1z5Tn , cl.exe seems to produce `??$emplace_back@H@?$vector@HV?$allocator@H@@@@QEAAAEAH$QEAH@Z` while clang-cl produces `??$emplace_back@H@?$vector@HV?$allocator@H@@@@QEAAAEAH$$QEAH@Z` (note `$` vs `$$`; the rest is identical).~~ Edit: This is incorrect; godbolt just shows it incorrectly...
@vincent-rg Thank you so much for fixing this! This has been really bothering me. I use `pdf` to test pdf changes on lagom often, and I ran into this multiple...
Isn't the ulimit for fds on Linux many thousands?
That's the soft limit though. Could `setrlimit` that up by 10x as a short-term fix :)
What's with the bots here? Looks like something real, somehow?