Thomas Munro <thomas.munro@gmail.com> writes:
> On Wed, Apr 10, 2024 at 5:03 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> I don't doubt that there are other clang versions where the problem
>> bites a lot harder. What result do you get from the test I tried
>> (turning mm_strdup into a no-op macro)?
> #define mm_strdup(x) (x) does this:
> Apple clang 15: master: 14s -> 9s
> MacPorts clang 16, master: 170s -> 10s
Wow. So (a) it's definitely worth doing something about this;
but (b) anything that would move the needle would probably require
significant refactoring of ecpg's string handling, and I doubt we
want to consider that post-feature-freeze. The earliest we could
land such a fix would be ~ July, if we follow past schedules.
The immediate question then is do we want to take Jelte's patch
as a way to ameliorate the pain meanwhile. I'm kind of down on
it, because AFAICS what would happen if you break the core
grammar is that (most likely) the failure would be reported
against ecpg first. That seems pretty confusing. However, maybe
it's tolerable as a short-term band-aid that we plan to revert later.
I would not commit the makefile side of the patch though, as
that creates the same problem for makefile users while providing
little benefit.
As for the longer-term fix, I'd be willing to have a go at
implementing the sketch I wrote last night; but I'm also happy
to defer to anyone else who's hot to work on this.
regards, tom lane