| 23 Apr 2023 |
 raitobezarius | it's the same as the ZFS issue | 13:24:38 |
| raitobezarius | BTF stuff | 13:24:39 |
 raboof | hmm, is that a recent(ish) thing? | 13:28:18 |
 Alyssa Ross | I don't think so | 13:28:49 |
| Alyssa Ross | at least, I don't think it's ever not been an issue for as long as we've had BTF | 13:29:06 |
| Alyssa Ross | but there was some recent upstream change that had us hopeful they might have fixed it | 13:30:29 |
| Alyssa Ross | https://lwn.net/Articles/928617/ | 13:30:33 |
| Alyssa Ross | hmm although re-reading it now, I don't think it will, because I don't think it was the order of struct members that was the problem | 13:31:15 |
| Alyssa Ross | I think it was the ordering of the structs themselves | 13:31:30 |
| raboof | https://r13y.com/ (june 2022) didn't seem to have problems, was that still on a pre-BTF kernel? | 13:32:28 |
| Alyssa Ross | probably | 13:32:55 |
| Alyssa Ross | we changed the default from 5.15 to 6.1 not that long ago | 13:33:24 |
 charuto | is r13y.com not supposed to be updated more regularly? | 13:36:49 |
| raitobezarius | its maintenance status is complicated, raboof has been doing a lot of work to fix that | 13:37:07 |
| raitobezarius | in reproducible.nixos.org | 13:37:11 |
| raboof | yeah, r13y.com its deprecated, we have (somewhat) more up-to-date reports at https://reproducible.nixos.org | 13:37:38 |
| charuto | thanks | 13:38:28 |
| charuto | sad to see so many regressions in reproducibility :( | 13:38:45 |
| raboof | I think some of them are just things we never noticed/hit before or that previously weren't part of the testset - but yeah the texlive and kernel changes seem regressions | 13:42:09 |
| raboof | the vim, terminus and grub ones all have PR's (the grub one even was just merged to staging) | 13:43:38 |
 @rnhmjoj:maxwell.ydns.eu | this is puzzling: why would these two files randomly disappear? | 13:52:18 |
| raitobezarius | rnhmjoj: look to the fix PR | 13:53:19 |
| raitobezarius | this is a parallel install problem | 13:53:25 |
| raitobezarius | make is unable to order the install targets required for one of the install target | 13:53:36 |
| charuto | In reply to @raboof:matrix.org
I think some of them are just things we never noticed/hit before or that previously weren't part of the testset - but yeah the texlive and kernel changes seem regressions oh, fair enough. hopefully we'll be soon able to see a 100% reproducible minimal iso again. | 13:57:52 |
| @rnhmjoj:maxwell.ydns.eu | In reply to @raitobezarius:matrix.org
make is unable to order the install targets required for one of the install target uhm, does that mean there's a prerequisite missing in some target? | 13:58:38 |
| raitobezarius | I thought it was a Make limitation | 14:01:38 |
| @rnhmjoj:maxwell.ydns.eu | This is the makefile:
install: $(PCF)
mkdir -p $(DESTDIR)$(x11dir)
for i in $(PCF) ; do gzip -c $$i > $(DESTDIR)$(x11dir)/$$i.gz ; done
fontdir:
mkfontscale $(DESTDIR)$(x11dir)
mkfontdir $(DESTDIR)$(x11dir)
fc-cache -f $(DESTDIR)$(x11dir)
This should be ideally fontdir: install, however you can't use a phony prerequisite, so it's a bi more complicated | 14:13:13 |
| @rnhmjoj:maxwell.ydns.eu | * This is the makefile:
install: $(PCF)
mkdir -p $(DESTDIR)$(x11dir)
for i in $(PCF) ; do gzip -c $$i > $(DESTDIR)$(x11dir)/$$i.gz ; done
fontdir:
mkfontscale $(DESTDIR)$(x11dir)
mkfontdir $(DESTDIR)$(x11dir)
fc-cache -f $(DESTDIR)$(x11dir)
This should be ideally fontdir: install, however you can't use a phony prerequisite, so it's a bit more complicated | 14:13:30 |
| raboof | In reply to @rnhmjoj:maxwell.ydns.eu
uhm, does that mean there's a prerequisite missing in some target? correct, creating a patch that adds all the installed outputs as requirements for the fontdir target would also make it deterministic, but I'm not sure if it's worth it tbh | 14:14:43 |
