IRC Logs for #crux-devel Thursday, 2009-06-04

*** sepen has joined #crux-devel01:34
*** jue has joined #crux-devel02:23
jueaon: without the patch I get this -> http://15b6ce51d9b71988.paste.se/02:24
*** sepen has quit IRC03:14
*** sepen has joined #crux-devel03:20
*** sepen has quit IRC03:48
*** sepen has joined #crux-devel03:49
*** jue has quit IRC04:00
*** jue has joined #crux-devel04:02
nipuLtilman: care to explain how this predicate/call back implementation works in pkgutils6?05:36
tilmancan you be more specific?05:41
tilmanPredicate just wraps some criteria that you want to check on a PkgPackage05:41
tilmanso has_package checks whether there's a package in the database that matches some criteria05:42
nipuLnever mind, it wasn't needed for what i wanted to do05:45
nipuLi want to make pkgrm more verbose about why it fails, so i can either check if the packge in installed first, or change pkg_database_remove to return int and define error codes05:48
nipuLthe first is easier, but probably less efficient05:48
tilmanplease define error codes05:48
nipuL#define ENOPKG 12305:49
tilmanwe should probably separate pkgutils' error codes from the standard ones05:50
tilmanso maybe PKG_PACKAGE_NOT_FOUND05:50
nipuLor we could paint the bike shed green05:51
tilmanhow is this bikeshedding?05:51
nipuLthe actual error code define name isn't overly iportant at this stage05:51
*** sepen has quit IRC06:03
nipuLsomething like? http://dpaste.com/51369/06:09
tilmanyes06:11
nipuLany plans on how to handle -r -u yet?06:37
tilmani think -r is already supported06:37
nipuLill look again06:37
tilmannipuL: pkg_database_new takes a const char *root06:37
tilmani was working on -u two weeks ago06:38
tilmannot finished ;)06:38
nipuLhmm, pkgadd gave an error when i used -r06:40
nipuLcould be the changes i made06:40
nipuLhow wuld -u be handled? remove the old package then install the new one?06:40
tilmanactually i didn't get that far. i wrote most of the conflict-detection stuff06:41
tilmanbut yes, i think that's how we'd do it06:41
tilmanmaybe we could also just overwrite stuff and remove the left overs afterwards06:42
nipuLlol, re error on -r; return (int) db_commit06:43
* nipuL slaps forhead06:43
tilmanhah06:43
nipuLthen there's all the pkgadd.conf hooks06:44
tilmanright :D06:44
nipuLwill be a good time to add some new ones06:44
nipuLINSTALL06:44
nipuLRUNSCRIPT06:45
tilmanINSTALL isn't new though06:45
nipuLhmm, if we have INSTALL hy go through all the bother of disabling nls?06:48
nipuLseems having INSTALL ^usr/share/locale/.*$ NO as default would be easier06:50
nipuLthen individuals could choose to keep their own locales if they want06:50
tilmantrue06:53
*** jtnl has joined #crux-devel07:23
*** mike_k has joined #crux-devel07:40
*** jtnl has quit IRC08:19
*** treach has joined #crux-devel08:23
tilmannipuL: i wondered about using INSTALL ^\.la$ NO08:30
*** sepen has joined #crux-devel08:33
nipuLthe absence of *.la would have made upgrading xlib/xcb recently easier08:33
tilmani don't think you gain anything by having them either08:34
tilmanafaik, libtool is seriously superfluous on linux08:34
nipuLi'm in favour of INSTALL hooks over manually removing in Pkgfiles08:35
tilmanwould locale files blow up the packages?08:35
tilmani guess they can be cmpressed nicely08:35
tilman.footprint is affected as well though08:36
*** sepen has quit IRC08:36
nipuLis that really a problem?08:37
tilmannot sure08:37
nipuLthe problem with such a change would be getting everyone on board08:37
tilmanwould be nice though -- the installer could ask "[ ] NLS-enabled crux"08:38
nipuL"[ ] Keep Documentation"08:39
tilman:]08:40
nipuLplaying with a lib for parsing pkgadd.conf, dynamic list or limit number of rules?08:43
tilmanlist08:44
tilmannipuL: git show 524844baf157901069e213bc87d28ec2af104ccf08:44
tilmanyou can git-revert that08:44
nipuLsweet08:45
nipuLshould use #warning "FIX ME" instead of /* FIX ME */09:06
tilmanthat breaks the build with -Werror09:09
*** sepen has joined #crux-devel09:36
*** jue has quit IRC09:38
*** jue has joined #crux-devel09:39
*** pedja has quit IRC09:51
nipuLanyhoo, time to sleep09:58
nipuLdid a little bit, http://git.die.net.au/cgit/crux/tools/pkgutils/?h=pkgutils6 if you want a peek09:59
*** sepen has quit IRC10:07
*** sepen has joined #crux-devel10:07
*** pedja has joined #crux-devel10:08
*** sepen has quit IRC11:02
tilmantux in french? tyx, of course :D11:41
*** sepen has joined #crux-devel11:48
*** fredg has joined #crux-devel11:59
*** fredg has left #crux-devel12:07
tilmanjue: do you know how to solve the vol_id problem with mdadm?12:11
tilmanjue: can i remove the dead symlink from udev in 2.6, or do i need to downgrade so mdadm can work?12:11
juetilman: I've committed 3.0 with a simple patch today12:12
juebut only for 2.612:12
jueand yes, I'd suggest to remove the symlink12:13
*** Rotwang has joined #crux-devel13:17
*** mike_k has quit IRC15:29
*** Rotwang has quit IRC15:31
*** treach has quit IRC16:30
*** sepen has quit IRC17:05
*** sepen has joined #crux-devel17:07
*** jue has quit IRC17:36
*** sepen has quit IRC18:07
nipuLugh, string handlling is such a pain in C19:33
nipuLtilman: I've put together a simple pkgadd.conf processor http://git.die.net.au/cgit/crux/tools/pkgutils/tree/lib/pkg_rules.c?h=pkgutils621:07
nipuLhaven't tested it yet though, will do that this afternoon21:11

Generated by irclog2html.py 2.11.0 by Marius Gedminas - find it at mg.pov.lt!