On 10/20 11:23, Theo Buehler wrote: > On Thu, Oct 13, 2022 at 06:07:51PM +0000, Jeremy Evans wrote: > > When I have time, I'll run a partial bulk of the PostgreSQL ports and > > try to fix any breakage. Hopefully I can complete that work by the > > release of PostgreSQL 15.1. > > There were three build failures: posgresql-pllua, repmgr, timescaledb. > Details below.
Theo, Thank you very much for running the bulk, that saves me a lot of time. With only 3 failures, the results are much better than in previous years. sthen@ committed a /etc/daily fix, so we can drop the new shared memory patch. > Given the non-trivial changes that are/were required in these consumers, > I would not be surprised by a variety of runtime fallout. Historically there hasn't been much runtime fallout from PostgreSQL upgrades. Hopefully by committing fairly early in the release cycle, we'll get sufficient testing before the release of OpenBSD 7.3. > databases/postgresql-pllua > -------------------------- > > Constrains on postgresql-client->=14,<15, postgresql-server->=14,<15. > Builds fine when bumping upper bounds to 16. > > See this, however: https://github.com/pllua/pllua/issues/10 Yep, we'll definitely want to patch that if they don't commit a fix upstream. > databases/repmgr > ---------------- > > repmgr-client.c:185:39: error: incomplete definition of type 'struct passwd' > strncpy(runtime_options.username, pw->pw_name, MAXLEN); > ~~^ > > postgresql/server/port.h no longer pulls in pwd.h. To fix build, > patching #include <pwd.h> into repmgr-client.c is enough. > > Upstream support for PostgreSQL 15 is only available from version 5.3.3 > https://repmgr.org/docs/current/release-5.3.3.html > This includes various 15-specific fixes (some relating to shared memory). We should upgrade to 5.3.3 to fix this. > databases/timescaledb > --------------------- > > CMake Error at CMakeLists.txt:362 (message): > TimescaleDB only supports PostgreSQL 12, 13 and 14 > > This is still the case with the latest timescaledb 2.8.1 and they still > seem to be sorting out fallout in the 2.9 branch. Pretty typical for timescaledb. I plan on marking it BROKEN when updating to PostgreSQL 15.1 if it isn't fixed by then. That's what we did during the PostgreSQL 13->14 upgrade. Thanks, Jeremy
