On 2026/11/01 13:55, Thomas Frohwein wrote: > On Sun, 1 Nov 2026 07:01:57 -0600 (CST) > Stuart Henderson <[email protected]> wrote: > > > CVSROOT: /cvs > > Module name: ports > > Changes by: [email protected] 2026/11/01 07:01:57 > > > > Modified files: > > infrastructure/mk: bsd.port.mk > > > > Log message: > > move _DT_WRKDIST before DISTNAME check, so that it can actually be > > used > > > > I've known _DT_WRKDIST to be inert for a while, but ports that combine > DIST_TUPLE with other ways to fetch their main DISTFILES and then > building in ${WRKDIR}/${DISTNAME} don't tolerate being overwritten by > _DT_WRKDIST. > > That's part of the reason why my proposed diff would rather get rid of > _DT_WRKDIST again [1]. > > I just tried building one of those ports, sysutils/py-tsk which > combines DIST_TUPLE with the MODPY universe and it breaks with this > commit and I can unbreak it with WRKDIST = ${WRKDIR}/${DISTNAME} > > Sorry, I don't think this is the right direction. Happy to be proven > wrong, but I've tried this unsuccessfully before. > > > [1] https://marc.info/?l=openbsd-ports&m=179221892747053&w=2
Ah yes my running bulk hit some issues with this that diffing pre/post-change "make show=WRKDIST" didn't show up, E=databases/arrow/cpp(i386-2) lang/node games/mvdsv(i386-1) astro/py-skyfield(i386-2)! devel/py-rx(i386-3) sysutils/firmware/bwfm(i386-1) (maybe others) Sorry, was already dealing with the already committed change that was going to cause issues for builds. Will try to investigate/fix the above tomorrow as long as my router with ixl doesn't fall over again.
