Hi,
On 2022-09-16 16:07:51 -0400, Tom Lane wrote:
> Andres Freund <andres@anarazel.de> writes:
> > On 2022-09-16 11:40:46 -0400, Tom Lane wrote:
> >> I suspect this is less about clang and more about LLVM APIs,
> >> but anyway it seems like we gotta fix something.
>
> > Yea, there's definitely a bunch of llvm 15 issues that need to be fixed - this
> > particular failure is pretty easy to fix, but there's some others that are
> > harder. They redesigned a fairly core part of the IR representation. Thomas
> > has a WIP fix, I think.
>
> I'm more and more getting the feeling that we're interfacing with LLVM
> at too low a level, because it seems like our code is constantly breaking.
> Do they just not have any stable API at all?
I don't think it's the wrong level. While LLVM has a subset of the API that's
supposed to be stable, and we mostly use only that subset, they've definitely
are breaking it more and more frequently. Based on my observation that's
because more and more of the development is done by google and facebook, which
internally use monorepos, and vendor LLVM - that kind of model makes API
changes much less of an issue. OTOH, the IR breakage (and a few prior related
ones) is about fixing a design issue they've been talking about fixing for 10+
years...
Greetings,
Andres Freund