jaeger changed the topic of #crux-devel to: CRUX (https://crux.nu/) development channel | Logs: https://libera.irclog.whitequark.org/crux-devel/
ivandi has quit [Quit: WeeChat 3.6]
ivandi has joined #crux-devel
jue has joined #crux-devel
<beerman> https://crux.nu/bugs/index.php?do=details&task_id=1852&project=1 <- next ticket. jue jaeger thoughts?
SiFuh has quit [Ping timeout: 260 seconds]
SiFuh has joined #crux-devel
<beerman> with farkuhar updating manuals, i guess we can also close this one? https://crux.nu/bugs/index.php?do=details&task_id=1201&project=1
SiFuh has quit [Ping timeout: 260 seconds]
SiFuh_ has joined #crux-devel
mnkydeth_ has joined #crux-devel
mnkydeth has quit [Ping timeout: 252 seconds]
<dlcusa> beerman, fine with me--I never understood why it stayed open, actually.
<beerman> dlcusa: well, a lot of things seem to stay untouched for far too long
<dlcusa> Ah. Tradition!
<beerman> I guess so ;)
<farkuhar> great to finally see these old tickets getting some attention. At the same time, we might try to avoid giving users a reason to open new tickets. cardinal seems to have stopped opening tickets after the one about polkit/mozjs78, although he could easily have followed up his emails about opt/babl with an actual Flyspray task.
<farkuhar> the fix for opt/babl was pretty simple: just cd into the top-level source directory before running meson setup. You can't run meson setup from the parent directory, because it spawns a git command without first checking for the existence of $PWD/.git
<beerman> babl has been updated to a new version which fixed the issue
jue has quit [Ping timeout: 260 seconds]
<farkuhar> that's good to hear. In theory I can appreciate SiFuh's stance that "all ports should be built on a core-only system" (if only to avoid footprint mismatches), but then you run the risk of not discovering "soft dependency" build failures until they get reported by the end users.
<beerman> you are welcome to provide a build farm to implement a proper CI system ;)
<beerman> while you are at it, why not rework prt-get and pkgmk so that it fully supports all optional dependency use cases and combinations
<beerman> you know that there used to be literally zero checks in Pkgfiles? at best, things were hard disabled
<farkuhar> Fun already proposed one such reworking of pkgmk, in FS#1576 (flexible footprint check). Most responses said his patch would violate the KISS principle, but the ticket remains open (cited reason: to provide a venue for further discussion of optional dependencies).
<beerman> there is no discussion when nobody is attending/contributing
<farkuhar> the Flyspray ticket itself gives the impression of very little discussion, but the IRC logs will tell a different story. Just a few weeks ago SiFuh was talking about making PKGMK_IGNORE_NEW=yes the default, almost as if it had already been decided.
<farkuhar> anyway, sorry for steering the discussion away from FS#1852. Regarding the check for id -u = 0, I don't see any harm in keeping it around. Prepending "sudo" or "doas" to any ports command that needs write permissions in a root-owned directory just feels like the more natural way to configure things, imho.
<SiFuh_> Are you talking about this comment, from the 24th of July? You should only ignore the footprint if it says NEW. If you ignore the footprint for MISSING your system will eventually break
<SiFuh_> Or this one from the 9th of August. Maybe I read wrong somewhere. I thought PKGMK_IGNORE_NEW would be defaulted to yes. Is it not happening now?
<farkuhar> Yes, your post from 9th August is what I meant.
<SiFuh_> I am sure that came up somewhere and I assumed that would be the default. But after testing 3.7 I realised it wasn't set.
<farkuhar> Re: FS#1852, it would be a more conservative change to replace the check for id -u = 0, rather than doing away with it entirely. Perhaps check whether the base directory of the ports tree is writable by the effective user? Therein lies the connection to FS#1111 (stop using a hard-coded base directory).
<SiFuh_> Might need to an else in there :-P
<beerman> the way I see it: IRC is a short term communication channel, while flyspray and our mailing lists should be long term discussion grounds.
<beerman> i still have a few unimportant personas filtered from my feed. there is also that one weird guy coming around every other year asking to drop glibc-32 for a pure 64 bit system
<beerman> and this is exactly the problem with irc, imo. discussion steer away naturally and then just end. i have been bringing up tickets every now and then just to get the queue cleaned up and have people know what we gonna do with whatever they proposed or reported.
mnkydeth__ has joined #crux-devel
mnkydeth_ has quit [Ping timeout: 268 seconds]