| From: | Thomas Munro <thomas(dot)munro(at)gmail(dot)com> |
|---|---|
| To: | pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: Requiring LLVM 14+ in PostgreSQL 18 |
| Date: | 2024-04-23 23:43:12 |
| Message-ID: | CA+hUKGLFNFMOPPME6mpkGeve0+6CRrZLMTKs+MmBGJp4vLcJ=A@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
Rebased over ca89db5f.
I looked into whether we could drop the "old pass manager" code
too[1]. Almost, but nope, even the C++ API lacks a way to set the
inline threshold before LLVM 16, so that would cause a regression.
Although we just hard-code the threshold to 512 with a comment that
sounds like it's pretty arbitrary, a change to the default (225?)
would be unjustifiable just for code cleanup. Oh well.
[1] https://github.com/macdice/postgres/commit/0d40abdf1feb75210c3a3d2a35e3d6146185974c
| Attachment | Content-Type | Size |
|---|---|---|
| v2-0001-jit-Require-at-least-LLVM-14-if-enabled.patch | application/octet-stream | 16.0 KB |
| v2-0002-jit-Use-opaque-pointers-in-all-supported-LLVM-ver.patch | application/octet-stream | 8.1 KB |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Imseih (AWS), Sami | 2024-04-24 01:40:45 | Re: query_id, pg_stat_activity, extended query protocol |
| Previous Message | David Steele | 2024-04-23 23:22:58 | Re: pg_combinebackup does not detect missing files |