| |
| |
| |
|
Page: 1 2
Comments:
<0> behdad: I figure I'll show up, we'll chat, then we'll drink beer. Is there another plan?
<1> keithp: what I have in mind is you running a session about fontconfig
<1> nothing that needs much preparation, if any
<0> and the beer? Is that on the schedule?
<1> sure it is. :)
<2> i think that starts at 0800
<0> Ok, I'm there.
<0> ajax: that seems awfully late; don't we usually start aroun 1700?
<1> keithp: we just need to formalize it a bit, cause some people need sponsorship, and OSDL needs a few paragraphs written, stuff, to think about it.
<1> :)
<1> keithp: ok, so marking you as confirmed
<0> yeah, if you'll do the blah-blah, we may be able to get more money. cool
<0> behdad: yes, I'm already marked as coming to the summit
<2> keithp: i guess it depends on whether 0800 < 1700 or not...
<1> I know. I created a subpage..
<1> 2000 PST is 1700 EST indeed
<1> keithp: so, fc-cache doesn't remove invalid caches?
<1> keithp: the error is a bit annoying in the first place, not talking about seeing it everytime
<0> behdad: it removes caches for directories which have disappeared
<0> behdad: otherwise, it leaves them around. Several people mentioned how they pre-cache lots of font directories and whack font paths around
<1> keithp: right, but invalid as in leftovers from previous version are not good
<0> oh, right, that's harsh
<0> Of course, it's possible that people are running multiple fontconfig versions at the same time
<0> which is why they use different filenames, btw
<1> keithp: then, just not print the error?
<0> oh, if it finds a cache file with the right name and wrong version?
<1> /home/behdad/.local/var/cache/fontconfig: invalid cache file: 81c87554abb5cbed44dc91927820b8dc.cache-2
<1> /home/behdad/.local/var/cache/fontconfig: invalid cache file: e166eff51bb88345c461ae8d1ca7510a.cache-2
<1> /home/behdad/.local/var/cache/fontconfig: invalid cache file: 0fc23c8259b45b7bffa09d2d2f5f79f0-x86.cache-2
<0> It should be removing those\
<1> the first couple don't have the right name even, right?
<0> That's when it prints the message, right before the unlink
<0> oh wait
<0> no, if it can't load the file, it skips it
<1> for a certain set I'm seeing them again and again
<0> It only removes files that it can identify as cache files that have the wrong version (or date)
<1> keithp: and the ones without the architecture suffix, why does it try them even?
<0> I decided caution was appropriate...
<0> it isn't that smart about filenames
<1> ok, I'll get back to you when bugs come in :)
<0> And, that particular form only existed between 2.3.95 and 2.3.96, so it should be rare..
<0> oh, you're running 2.3.95 aren't you.
<1> normally yes, in fedora, unfotunately no :(
<1> yes
<1> we ship it!
<0> That's insane, you know
<1> ok, /me continues the push
<0> behdad: you should be able to clean those out in your rpm spec file
<1> pre-release is a very relative term
<0> behdad: you might check what I push into debian :-)
<0> debian is currently running 2.3.2
<1> well, FC5 has 2.3.94
<1> so, technically 2.3.95 has not been in stable releases
<1> debian is a lot more conservative than fedora of course
<0> behdad: 2.3.2 is in debian *unstable*
<0> I would have pushed something more recent if I thought it was a good idea
<1> what's this:
<1> Working on: /home/behdad/src/redhat/fontconfig/devel/fontconfig-2.3.97/fc-cache/../fc-cache/fc-cache.sgml
<1> <joss@debian.org>
<1> JosselinMouette2003Josselin MouetteSep 22, 2005Done.
<1> rm -f manpage.refs manpage.links
<1> keithp: well, we ***umed that you are not paying much attention to fontconfig development
<0> behdad: true in too many senses
<1> (and it wasn't me who was in charge btw)
<1> anyway, I have a weird linking problem here
<1> fc-cache is linking to my installed fontconfig, not the one being built
<0> in any case, dealing with the 2.3.94 -> 2.4 transition is not that huge a problem, and should be scriptable in a reasonably efficient fashion
<0> ouch. With the .so version change, that 'shouldn't happen'.
<1> oh, this may be the libtool issue I've known
<0> heh. libtool considered evil
<0> behdad: I think I will push davidt's optimization; it seems obvious, and generated matching cache file contents
<1> I ***ume all but major parts of the so version are informational
<1> ok, I'm keeping it here in fedora for now
<0> no, the other versions select preferred ones for use; ldconfig sets all that up
<0> behdad: no, I mean, I'm going to push it to master so it ships in 2.4
<1> keithp: I know.
<1> and I said that I'll keep it in my package then (instead of dropping it)
<0> ah, ok
<1> keithp: right about ldconfig, but that doesn't help in linking against an old version when a new enough one is not available
<0> oh, true enough
<0> behdad: and sed is broken on OS X. Shocking.
<1> sigh
<0> that's ok, I found a nicer way to do it anyway
<0> and found another bug in the public header along the way
<1> cool
<1> keithp: can you make generated changelogs more verbose?
<0> behdad: they have the whole commit message
<1> keithp: freel free to checkout / copy / adjust cairo/ChangeLog.mk
<1> keithp: in cairo we include file change stats now
<0> oh, that'd be cool
<1> people where complaining about them missing compared to CVS
<0> yeah
<0> ok
<0> Ok to relicense under MIT license?
<1> and the stuff in cairo caches git log output as that can take quite some time with the stats enabled
<1> keithp: Ok for you about anything I write, Ok? :)
<0> behdad: tnx. I'll still probably ask every time :-)
<1> :)
<0> behdad: your employer is fairly 'everything should be GPL', so I'd hate to run afoul of them
<0> behdad: there's lots of make-fu to build these scripts; any of use?
<1> keithp: most of the more interesting stuff I write is not paid by my employer :)
<1> keithp: well it's a bit bound to cairo's versioning scheme
<0> yeah, not so good for me
<1> here is the interesting part for you:
<0> I'll figure it out
<1> basically the $(srcdir)/ChangeLog: target
<0> I think I just need to nuke the version stuff
<1> sans prev,nearest_tag stuff
<1> the caching is cool though
<1> it caches ChangeLog until the latest tag you have, as in output of git-describe
<0> cool
<0> can you suggest a replacement for PREV_CHANGELOG_VERSION? That part appears magic
<1> keithp: empty I suppose, if you want to keep a single ChangeLog
<1> keithp: or rather, your init tag
<1> the first commit in the repo
<0> I like to have both changelogs, like cairo
<0> note that I stick the diff changelog in the download directory
<0> behdad: if you want to adapt your change log fu to fontconfig, that would be cool
<1> ok, sure.
<1> so, 2.3.97 didn't build on ppc64
<1> keithp: ./fc-arch: unknown signature "12345678_00000001_00000008_00000004_00000008_00000018_00000008_00000008_00000010_00000004_00000008_00000010_00000004_00000008_00000018_00000008_00000008_00000002_00000020_00000004_00000030" Please update fcarch.tmpl.h and rebuild
<1> keithp: I thought you added them
<1> fc-arch segfaults...
<0> behdad: cool!
<1> keithp: fc-arch segfaults if no arch p***ed on commandline
<0> behdad: yeah, big deal
<1> keithp: so, should I call that ppc64-4k?
<0> behdad: no, no page size dependence
<1> or rather, ppc64-32?
<0> but, I did add ppc
<0> just ppc64
<1> as in "int is 32"
<1> ok
<0> behdad: can I get an rsync feed for gnome CVS somehow?
<1> keithp: if you ask on gnome-infrastructure list I guess
<0> behdad: regular ssh access to cvs.gnome.org would suffice
<1> or maybe not for CVS, not sure.
<0> I just need rsync to fetch the repository
<0> nasty of them not to permit that
<1> donno
<1> they surely do
<1> Owen is the master there
<3> I can't get composite to load
<3> fixes/damage are loading but composite isn't
<4> what error
<4> you enabled the extension?
<4> Section "Extensions" Option "Composite" "Enable" Option "RENDER" "Enable" Option "DAMAGE" "Enable" EndSection ?
Return to
#freedesktop
or
Go to some related
logs:
nvtvout xorg
#math
suse10.0 pakages
webmarshal debian
userspace firmware loading support
burn kdvd
fc5 tree connect failed: NT_STATUS_BAD_NETWORK_NAME
uploadProgress Shadda
#perl
lvm shrink centos
|
|