previous
2019-08-08
12:58:34 <rrq> just look through the ganeti1 hosts, and though some might be large enough I couldn't find any iso building...
13:00:45 <rrq> hmm maybe files.devuan.org ..
13:07:27 <rrq> (nope; not building -- just storing)
18:05:04 <fsmithred> From KatolaZ: Both jessie (since beta2) and ascii (since the first beta) were built
18:05:05 <fsmithred> on a dyne server (naboo). Just ask jaromil/parazyd. There should be a
18:05:05 <fsmithred> folder under my home there with a README and all the steps needed to
18:05:05 <fsmithred> build the installer ISOs.
---------- 2019-08-09 ----------
07:29:59 <rrq> jaromil: parazyd: is access to "naboo" on option for mortals?
07:33:28 <rrq> fsmithred: I'm trialling nuilding mini.iso with debian-installer, and ends up with a grub2 flaw in the installation test...
07:34:50 <rrq> I'm guessing you've been there before me...
07:35:41 <rrq> (the grub-install bug of inventing .gmo files that it then is so unhappy about that it fails)
07:35:58 <rrq> (or something)
07:36:40 <rrq> is this something you know something about?
08:05:15 <rrq> ah; the "bad" locale actually are those from consolekit=1.2.1-4
08:06:35 <golinux> https://dev1galaxy.org/znclog/search.cgi?tz=-900&P=%23devuan&all=y&q=gmo
08:11:00 <rrq> yes. that bug. it concerns exactly the 4 *.po files of the last commit in gitlab for devuan-packages/consolekit2
08:12:23 <rrq> probably due to the spurious copy-pasted newline (eg line 17 in po/es.po)
08:13:47 <rrq> though I'm not really familiar with that "po" language so I'll make a bug report rather
08:27:32 <fsmithred> rrq, I think I ran into that problem once, and the fix had nothing to do with the locale files
08:31:25 <rrq> well, those 4 locales certainly cause the grub-install failure for me right now, and removing them one by one gives successive failure pointing at the next one, until they are all removed.
08:34:01 <fsmithred> removing them from /usr/share/?
08:34:21 <rrq> from /usr/share/locale/
08:34:52 <rrq> on the image being installed
08:36:16 <rrq> unfortunately it's a long chain from fix to test, so I'll file a bug report first
08:37:00 <rrq> then work out a way to "lab test" it
08:42:14 <fsmithred> I can't find any .gmo files
08:43:43 <rrq> no, not after a successful installation; these are on the install image when grub-install breaks
08:45:06 <rrq> you'll have to run a new installation up to the breaking grub-install, then use c-a-f2 to review
08:47:02 <fsmithred> I just checked in three installations, jessie, ascii and beowulf, point being that none of the programs installed use .gmo files
08:47:11 <fsmithred> they all use .mo files
08:48:13 <fsmithred> so is the bug the ck makes .gmo files or that grub doesn't recognize them?
08:50:30 <rrq> afaict, consolekit provides buggy .po files that gettext invoked by grub-install fails on, leaving .gmo directories that then grub-install gets upset about
08:51:37 <rrq> (refuses to copy directory onto /boot)
08:51:58 <fsmithred> I might be able to find my notes on how to check a po file
08:59:36 <fsmithred> Nope. I might just be thinking of xgettext error messages.
2019-08-09
next