From mboxrd@z Thu Jan 1 00:00:00 1970 From: Danny Milosavljevic Subject: Re: hplip package (work in progress) Date: Sun, 20 Mar 2016 01:19:04 +0100 Message-ID: <20160320011904.50131806@scratchpost.org> References: <20160312191233.48a661ca@scratchpost.org> <8737ruuj0l.fsf@gnu.org> <20160315210512.23989245@scratchpost.org> <20160315221931.16abe3dc@scratchpost.org> <87vb4mygw4.fsf@igalia.com> <87mvpyyfon.fsf@igalia.com> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Return-path: Received: from eggs.gnu.org ([2001:4830:134:3::10]:35192) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ahR5V-0004GA-Os for guix-devel@gnu.org; Sat, 19 Mar 2016 20:19:23 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ahR5S-0007Zn-IM for guix-devel@gnu.org; Sat, 19 Mar 2016 20:19:21 -0400 Received: from dd1012.kasserver.com ([85.13.128.8]:33260) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ahR5R-0007ZW-Dm for guix-devel@gnu.org; Sat, 19 Mar 2016 20:19:18 -0400 In-Reply-To: <87mvpyyfon.fsf@igalia.com> List-Id: "Development of GNU Guix and the GNU System distribution." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: guix-devel-bounces+gcggd-guix-devel=m.gmane.org@gnu.org Sender: guix-devel-bounces+gcggd-guix-devel=m.gmane.org@gnu.org To: Andy Wingo Cc: guix-devel@gnu.org Hi, On Wed, 16 Mar 2016 09:45:12 +0100 Andy Wingo wrote: > > Generally, yes. If it later needs to be merged into a common /etc/cups > ^, put it in the hplip store directory > > directory for some reason, then cups-service needs to create a union > > directory based on the etc/cups directories in a set of packages. See > > polkit-service for an example. It's trying to update mime.types of cups in order to add pstotiff (pstotiff.convs and pstotiff.types) to it. hplip's "configure" checks whether there is /usr/share/cups/mime/mime.types . If there is, it prefers that. Otherwise, it tries to use /etc/cups . It can be overridden by configure-flag --with-mimedir=... or disabled by --disable-fax-build However, I think in this case we have to merge the current system's mime.types together. Checking for mime-types in guix, I found - ruby mime types - perl mime types - cups mime types - haskell mime types I guess in this case we should use cups mime types. cups.scm fiddles around with cups mime types for some tests and also for installation ('install-cups-filters-symlinks). $out/share/cups/mime/cupsfilters.types and $out/share/cups/mime/cupsfilters.convs and $out/share/cups/drv/cupsfilters.drv and $out/share/ppd . I think all of these need to do some merging. Cups already does symlink the mime config files from package "cups-filters" into its output dir - but it does not merge. I also checked polkit-service and it seems to use something called "service extensions" in order to concatenate all the different actions. It seems to use polkit-configuration-actions (which is always empty? - how does this get set?) I'd like to ask for help on this, I have no idea how this part works in Guix. Then, after all this, hplip has hardcoded inside Makefile.am : # hpmud.rules rulesdir = /etc/udev/rules.d if UDEV_SYSFS_RULES dist_rules_DATA = data/rules/56-hpmud_sysfs.rules else dist_rules_DATA = data/rules/56-hpmud.rules endif halpredir = /usr/share/hal/fdi/preprobe/10osvendor dist_halpre_DATA = data/rules/20-hplip-devices.fdi # hplip.conf hplip_confdir = /etc/hp hplip_conf_DATA = hplip.conf # hplip.state hplip_statedir = /var/lib/hp dist_hplip_state_DATA = Should we substitute those? (I'm already autogen'ing, so it doesn't matter to substitute some more things) What about /etc/hp/hplip.conf ? Should it be autogenerated (or should we let hp-setup write it?) and we just put the distfile into the store as an example? I don't think we use hal, so I'll just let it install a useless file, if it absolutely has to. Lirc seems to do the same. It seems the udev files are already being merged by services/base , right? I tried to find the generated udev.conf files, and there is for example (I digress): 50-udev-default.rules:KERNEL=="fuse", MODE="0666", OPTIONS+="static_node=fuse" 99-fuse.rules:KERNEL=="fuse", MODE="0666" I think that's why fuse autoloading doesn't work anymore.