**** BEGIN LOGGING AT Sat Feb 05 19:33:03 2005
Feb 05 19:37:32 <VoodooZ_home> me too. 
Feb 05 19:37:40 <VoodooZ_home> but it sure took forever.
Feb 05 19:37:52 <VoodooZ_home> I'm timing my build to see how much of a diff it makes.
Feb 05 19:38:25 <jacques> unfortunately it didn't fix my oe build problem
Feb 05 19:38:34 <VoodooZ_home> what problem is that again?
Feb 05 19:39:58 <jacques> i seem to be the only person who has it
Feb 05 19:40:08 <VoodooZ_home> it seems to have fixed mine anyways. libtool completed. 
Feb 05 19:40:14 <VoodooZ_home> but it's not over yet!
Feb 05 19:40:14 <jacques> cool
Feb 05 19:40:36 <VoodooZ_home> What's your problem?
Feb 05 19:40:45 <VoodooZ_home> (with OE)
Feb 05 19:40:55 <ByronT> well that was not fun
Feb 05 19:41:27 <jacques> quilt isn't working right
Feb 05 19:42:37 <VoodooZ_home> really? I had some problems with that originally when they started using it. 
Feb 05 19:42:58 <jacques> one one machine it works fine, on the other it always fails
Feb 05 19:43:11 <jacques> chmod: cannot access `patches/*': No such file or directory
Feb 05 19:43:19 <jacques> cp: cannot create regular file `patches/fix-ipkg-link': Permission denied
Feb 05 19:43:47 <VoodooZ_home> odd.
Feb 05 19:44:21 <VoodooZ_home> I noticed i had permission denied errors when wiping my tmp folder lately. It forces me to become root to do it.
Feb 05 19:44:37 <VoodooZ_home> Are the perms on those folder/files the same?
Feb 05 19:45:15 <jacques> the problem is that the patches dir is never created
Feb 05 19:46:27 <VoodooZ_home> did you mention it to kergoth?
Feb 05 19:46:37 <jacques> yes
Feb 05 19:48:00 <VoodooZ_home> did he have any leads?
Feb 05 19:48:06 <jacques> no
Feb 05 19:48:27 <VoodooZ_home> damn. that sucks when even kergoth doesn't have a clue.
Feb 05 19:48:45 <jacques> that machine's down for the count. every week or so I clear all the caches, install the latest bb, etc, and try again
Feb 05 19:48:48 <VoodooZ_home> I would try reinstalling the whole thing. 
Feb 05 19:48:57 <jacques> the whole what thing?
Feb 05 19:49:06 <jacques> I already recloned oe
Feb 05 19:49:10 <VoodooZ_home> I meant the OE stuff and related apps.
Feb 05 19:49:16 <VoodooZ_home> but I'm sure you alrady did that
Feb 05 19:49:19 <jacques> redownloaded quilt cvs about 10 times
Feb 05 19:50:12 <jacques> totally cleared sources dir
Feb 05 19:50:53 <VoodooZ_home> There has to be a reason. Perhaps you can try to isolate the file that calls quilt and then put some debug echos everywhere .
Feb 05 19:51:13 <VoodooZ_home> Then again, I have no clue what I'm talking about.
Feb 05 20:03:22 <CIA-7> 03garpinc * 10unslung/ (3 files in 2 dirs): OpenNap Plugin for giFT
Feb 05 20:13:43 <kergoth`bbl> jacques: if you cant get quilt working on that machien with oe, dont use quilt. inherit the bbclass that switches you from quilt to patcher.
**** ENDING LOGGING AT Sat Feb 05 20:14:00 2005

**** BEGIN LOGGING AT Sat Feb 05 20:15:04 2005
Feb 05 20:15:29 <CIA-7> 03garpinc * 10unslung/Makefile: Added gift-opennap for testing
Feb 05 20:41:53 <CIA-7> 03djf * 10unslung/Makefile: added gettext to NATIVE_AND_CROSS_PACKAGES_READY_FOR_TESTING
Feb 05 21:02:26 <VoodooZ_home> wow! My "fat" openslug build now completes in 74minutes!! 
Feb 05 21:03:44 <VoodooZ_home> NOT having the ccache folder on an nfs drive (by accident) helps! ;)
Feb 05 21:05:05 <ka6sox-away> hey randy
Feb 05 21:05:29 <randman> hi, thanks
Feb 05 21:10:11 <ByronT> heading to bed... later all
Feb 05 21:28:21 <CIA-7> 03djf * 10unslung/make/irssi.mk: initial checkin - does not build yet - working on it
Feb 05 22:00:18 <VoodooZ_home> Anybody seen that boot messge with the newer openslug? : INIT: VERBOSE=no requested, but /proc/progress not available.
Feb 05 22:30:22 <rwhitby-away> another $25 just rolled in ....
Feb 05 22:42:34 <kergoth`bbl> VoodooZ_home: that likely has to do with the work mickeyl has done with the linux progress patch stuff.  ask him.
Feb 05 22:45:41 <VoodooZ_home> ok. thanks.
Feb 05 22:46:15 <VoodooZ_home> progess patch? 
Feb 05 22:47:29 <kergoth`bbl> ~lpp
Feb 05 22:47:30 <jbot> it has been said that lpp is The Linux Progress Patch. It gives you graphical boot screens while booting the kernel. Like MSWindows. http://lpp.freelords.org/
Feb 05 22:47:48 <jacques> hmm
Feb 05 22:47:49 <kergoth`bbl> he was playing around with that or a related project on his zaurus running 2.6
Feb 05 22:47:55 <kergoth`bbl> dont recall the details
Feb 05 22:52:28 <VoodooZ_home> cool. thanks. I'll look it up.
Feb 05 22:53:48 <VoodooZ_home> I also discovered a new problem with udev under openslug.It seems the links.conf created extra nodes (leds and buzzer are not detected automatically) never make it.
Feb 05 22:54:22 <VoodooZ_home> when I manuall restart udev after logging it though it creates them no problem. as if the ramfs is mounted on top of them or something.
Feb 05 22:55:15 <kergoth`bbl> ramfs is mounted before udevstart gets run
Feb 05 22:55:18 * kergoth`bbl shrugs
Feb 05 22:55:47 <VoodooZ_home> I know. but I can't seem to see why they never get created like before. Oh well.
Feb 05 22:56:02 <VoodooZ_home> I might remove udev anyways. It doesn't give me anything more really.
Feb 05 22:57:39 <VoodooZ_home> for something with more or less fixed hardware like a robot there's no big advantage.
Feb 05 22:57:48 <VoodooZ_home> anyways. good night. It's 2am here so...
Feb 05 22:58:33 <kergoth`bbl> nod, its more useful when you're actually hotplugging things.
Feb 05 22:59:11 <VoodooZ_home> yeah. It might happen with the bot but rarely so we'll see.
Feb 05 22:59:14 <VoodooZ_home> take care,
Feb 05 23:30:45 <rwhitby-away> hmm - pkgconfig fails on cross on builds ...
Feb 05 23:31:04 <rwhitby-away> Can't exec "aclocal": No such file or directory at /usr/bin/autoreconf line 176.
Feb 05 23:32:19 <rwhitby-away> ah - builds was missing automake
Feb 05 23:47:47 <CIA-7> 03rwhitby * 10unslung/Makefile: Promoted gettext and pkgconfig
Feb 05 23:50:31 <fontenot> hi ho
Feb 05 23:50:37 <rwhitby-away> hmm - maybe gettext is failing native
Feb 05 23:50:38 <fontenot> irssi on slug
Feb 05 23:50:47 <rwhitby-away> jacques: how did you get gettext to compile native?
Feb 05 23:50:54 <fontenot> hmm? shouldn't - what error?
Feb 05 23:51:16 <rwhitby-away> grep: /home/slug/packages/toolchain/armv5b-softfloat-linux/gcc-3.3.5-glibc-2.2.5/armv5b-softfloat-linux/lib/./libstdc++.la: No such file or directory
Feb 05 23:51:16 <rwhitby-away> libtool: link: `/home/slug/packages/toolchain/armv5b-softfloat-linux/gcc-3.3.5-glibc-2.2.5/armv5b-softfloat-linux/lib/./libstdc++.la' is not a valid libtool archive
Feb 05 23:51:16 <rwhitby-away> make[5]: *** [libasprintf.la] Error 1
Feb 05 23:51:36 <fontenot> yeah you have .la files somewhere
Feb 05 23:51:44 <fontenot> .la files with those paths in them
Feb 05 23:51:56 <fontenot> either in /opt/lib or in the toolchain lib dir
Feb 05 23:52:35 <fontenot> those can cause build problems for more than just gettext
Feb 05 23:54:06 <fontenot> for now at least it looks like we will have another native-only - irssi - don't try it yet tho
Feb 05 23:54:23 <fontenot> do you think ppl will want ipv6 support in irssi?
Feb 05 23:54:40 <ka6sox> is anybody actually using it?
Feb 05 23:57:52 <jacques> ipv6 ?
Feb 05 23:58:11 <jacques> freenode supports it. I've seen lots of ppl signed into freenode with ipv6 addresses
Feb 06 00:06:12 <rwhitby-away> jacques: any idea which packages would be erroneously installing those .la's ?  And how come you don't have them - another manual change which hasn't been reflected in an ipk correction yet?
Feb 06 00:09:06 <CIA-7> 03djf * 10unslung/sources/irssi/control: initial checkin
Feb 06 00:10:25 <jacques> rwhitby-away, I remove some manually, and others I have the makefile remove when it makes the ipk
Feb 06 00:10:34 <jacques> and some dont seem to cause any problems
Feb 06 00:10:58 <jacques> and pretty much any package which installs libs and uses libtool will install .la's
Feb 06 00:11:35 <jacques> the libstdc++ ones seem to cause the most problems, but I have had problems with others
Feb 06 00:31:02 <dyoung> re
Feb 06 00:31:23 <rwhitby-away> so should we outlaw .a's and .la's from /opt/lib ?
Feb 06 00:32:49 <rwhitby-away> hmm - pkgconfig fails cross too
Feb 06 00:32:59 <rwhitby-away> strip: Unable to recognise the format of the input file /home/unslung/packages/builds/pkgconfig-0.15.0-ipk/opt/bin/pkg-config
Feb 06 00:33:32 <jacques> i think it failed much earlier than tat for me on cross
Feb 06 00:33:43 <jacques> that's an easy fix
Feb 06 00:34:05 <jacques> it just means stupid install is trying to use the host strip instead of cross-strip
Feb 06 00:34:25 <jacques> for me cross failed on some auto* stuff
Feb 06 00:34:28 <jacques> native worked tho
Feb 06 00:35:01 <jacques> gentoo installs all known versions of autoconf and automake and then uses a wrapper with env vars to select between them
Feb 06 00:35:22 <jacques> since they are not backward nor forward compatible (yay auto* devs)
Feb 06 00:45:07 <rwhitby-away> so should we outlaw .a's and .la's from /opt/lib ?
Feb 06 00:49:14 <jacques> .la's : for now, I would say yes - we aren't deriving any benefits from them as long as them have incorrect info in them
Feb 06 00:49:45 <jacques> .a's : where a .so exists I would say yes, but some packages on'y make .a's (I think I noticed some like this anyway)
Feb 06 00:50:15 <jacques> I think libpcap is an example of a package that only makes a static lib
Feb 06 00:50:35 <rwhitby-away> we shouldn't be putting .a's in /opt/lib - we want shared libraries on a small memory slug
Feb 06 00:51:01 <rwhitby-away> if a package needs a .a, it can get it from staging.
Feb 06 00:51:11 <jacques> ok 
Feb 06 00:51:27 <rwhitby-away> bbiab
Feb 06 01:01:19 <dyoung> I merged the 3.18 unslung-standard-rootfs into 4.x bumping the 4.x PR to 50.  (from 41)
Feb 06 01:10:13 <CIA-7> 03djf * 10unslung/make/irssi.mk: builds native, not cross - ready for testing
Feb 06 02:01:06 <CIA-7> 03rwhitby * 10unslung/Makefile: Moved gettext and pkgconfig to cross only.
Feb 06 02:32:33 <CIA-7> 03rwhitby * 10unslung/Makefile: Demoted pkgconfig to NEEDS_TO_BE_FIXED.
Feb 06 02:33:28 <rwhitby-away> reports of pkgconfig working were greatly exaggerated :-)
Feb 06 02:33:38 <jacques> no they arent
Feb 06 02:33:44 <jacques> I said it worked for me native and it did
Feb 06 02:33:50 <jacques> that's not exagerration
Feb 06 02:34:15 <rwhitby-away> not disputing that, but it doesn't work on a standard native compilation environment (as per the wiki page)
Feb 06 02:34:40 <rwhitby-away> so your success can't be reproduced automatically in a regular build
Feb 06 02:35:27 <jacques> it would help if I knew what the error was
Feb 06 02:36:23 <rwhitby-away> jacques: no need to get defensive - it's the .la error.  It's not pkgconfig's fault, it's all the other packages that install .la's.  But it doesn't change the fact that I can't automatically build it right now, hence it gets demoted.
Feb 06 02:37:52 <jacques> so what's the plan? leave the affected packages out of the builds until "someone" fixes all the packages that install .la's ?
Feb 06 02:38:40 <rwhitby-away> yep
Feb 06 02:39:17 <rwhitby-away> unfortunately, the last package looses until the others are fixed, cause I can't really remove packages from the feed which have been working for people for weeks
Feb 06 02:39:32 <rwhitby-away> so the one transitioning from testing draws the short straw
Feb 06 02:39:54 <rwhitby-away> also for pkgconfig cross it is trying to strip a non-binary.
Feb 06 02:40:02 <rwhitby-away> (as I reported earlier)
Feb 06 02:40:32 <jacques> yeah and it doesn't get nearly that far cross for me, for the reasons I gave
Feb 06 02:40:56 <rwhitby-away> I have to wonder if jp30 tested pkgconfig-ipk, or whether he figured that just getting a binary is enough ...
Feb 06 02:41:36 <jacques> I dunno, but lack of pkgconfig is going to block a lot of stuff
Feb 06 02:41:40 <jacques> like irssi
Feb 06 02:42:14 <rwhitby-away> well, getting it to work cross should be easy if it's just a strip error when cross-compiled on builds.
Feb 06 02:42:50 <jacques> it just needs to be explicitly stripped, not install --strip or install-strip whatever it's doing now
Feb 06 02:42:57 <rwhitby-away> agreed
Feb 06 02:43:06 <dyoung> DIdnt I mention that a few months ago?
Feb 06 02:43:25 <dyoung> something along the lines of "install-strip is broke, dont do that" ?
Feb 06 02:43:57 <jacques> it's not always broken, some packages use it and it works - install --strip (or -s I always forget) will always be broken on cross
Feb 06 02:43:57 <rwhitby-away> yep, but we have new developers every day, so unless it's on the packaging best practises page, then it gets forgotten
Feb 06 02:44:01 <dyoung> which is why a bunch of mk files I touched are ugly in that they are stripping millions of binaries.
Feb 06 02:44:32 <rwhitby-away> http://www.nslu2-linux.org/wiki/Unslung/PackagingBestPractices
Feb 06 02:44:40 <jacques> it really depentds how the install-strip target is defined in the individual Makefile
Feb 06 02:45:17 <jacques> but I think install [-s, --strip] will always fail
Feb 06 02:45:28 * dyoung nod
Feb 06 02:45:51 <dyoung> I was making a over generalized observation because almost everything I touched didnt work. 
Feb 06 02:46:13 <dyoung> not that way it was expected to at least.
Feb 06 02:46:29 <rwhitby-away> can we fix install instead?
Feb 06 02:46:33 <jacques> for a large package with a properly written Makefile, install-strip can save a lot of work, which is why I'm trying to avoid making a rule against using it
Feb 06 02:46:59 <dyoung> I think it should be up to the packager to decide whether or not it works.
Feb 06 02:47:35 <rwhitby-away> well, it seems that pkgconfig is an automake generated Makefile, and it does:
Feb 06 02:47:37 <jacques> I see nothing in the install man page to tell it to use anything other than the host strip command
Feb 06 02:47:44 <rwhitby-away> install-strip:
Feb 06 02:47:44 <rwhitby-away>         $(MAKE) $(AM_MAKEFLAGS) AM_INSTALL_PROGRAM_FLAGS=-s install
Feb 06 02:47:52 <jacques> ouch
Feb 06 02:48:03 <rwhitby-away> so you're saying that automake generated makefiles will always fail install-strip
Feb 06 02:48:10 <jacques> who said that?
Feb 06 02:48:38 <rwhitby-away> s/you're/we're/
Feb 06 02:48:47 <jacques> I don't think so
Feb 06 02:48:51 <dyoung> I have to wonder if thats automake version dependant.
Feb 06 02:49:00 <dyoung> jacques, what version of automake do you use usually?
Feb 06 02:49:01 <jacques> it's nearly impossible to make generalizations about auto* generated stuff
Feb 06 02:49:05 <rwhitby-away> # Makefile.in generated automatically by automake 1.4-p4 from Makefile.am
Feb 06 02:49:25 <dyoung> I'm generally using 2.59
Feb 06 02:49:30 <rwhitby-away> Makefile.am contains nothing special about strip
Feb 06 02:49:40 <jacques> because of all the different versions and the various ways they can be used in the .am/.ac/.in whatever files
Feb 06 02:49:49 <dyoung> And in general it doesnt do the right thing for me.  But I might not be licking it right either.
Feb 06 02:50:27 <jacques> because the auto* stuff is so screwed up and not compatible with itself, gentoo has to install all known versions and uses a wrapper which chooses versions based on an env var
Feb 06 02:50:54 <jacques> so I have about 5 versions of autoconf and 6 of automake installed on my gentoo systems
Feb 06 02:51:04 * dyoung head spins
Feb 06 02:51:30 <rwhitby-away> as far as I can see, the version of autoconf and automake on builds.nslu2-linux.org, with the most basic Makefile.am (with no specific stuff for stripping) will always generated a Makefile with a install-strip target which fails on cross compiles.
Feb 06 02:51:35 <jacques> if you have a gentoo system, look at the .ebuild's for those sometime - scary
Feb 06 02:53:27 <dyoung> 2.59 is on builds.
Feb 06 02:53:40 <jacques> I find it disgusting that the auto* stuff, which is supposed to make cross-compiling easier, would do that
Feb 06 02:54:01 <jacques> but not in the least bit surprising, since I have used it for a long time
Feb 06 02:54:24 <dyoung> I ran into 2.older not working for some things, and it told me something like "hey upgrade to 2.59 to make this work!"
Feb 06 02:54:39 <jacques> yeah the them soma packages only work with the older versions
Feb 06 02:54:49 <jacques> hence the gentoo approach
Feb 06 02:55:22 <jacques> not too long ago it was a stupid game, upgrade autoconf to build pacakge A, now package B no longer builds
Feb 06 02:55:57 <jacques> like libtool, sometimes it seems to cause more problems than it's worth
Feb 06 02:56:37 <dyoung> I ran into that .la / .a stuff too.  So I was generally blowing them away.
Feb 06 02:57:17 <dyoung> Libtool just seems evil.
Feb 06 02:57:24 <jacques> unless we get a libtool expert, I don;t see us having any other way to deal with it
Feb 06 02:57:46 <jacques> any package built against libs in staging (cross or native) will have bad info in the .la's AFAICT
Feb 06 02:58:03 <dyoung> nod
Feb 06 02:58:04 <jacques> only ones that dont are built native against libs in their final installed locations
Feb 06 02:58:24 <jacques> now, I suppose we could edit all the .la's, but currently I don't see the point
Feb 06 02:58:35 <jacques> I would have to understand the benefits of the .la's for that
Feb 06 02:58:58 <dyoung> quicker parsing maybe?
Feb 06 02:59:09 <dyoung> Dunno...
Feb 06 02:59:22 <jacques> some fuzzy thing about being able to find libs that other libs need or something
Feb 06 02:59:46 <dyoung> How is this handled for OE built packages?
Feb 06 02:59:59 <jacques> and all the build processes warn about needing to do libtool --finish or something which we never do
Feb 06 03:00:07 <jacques> that would have to be done in postinst
Feb 06 03:00:25 <jacques> dyoung that would be useful to know
Feb 06 03:00:56 <jacques> oh and ever do man libtool ?
Feb 06 03:01:23 <jacques> none of my systems have a man page for it
Feb 06 03:01:33 <dyoung> theres a manpage on builds
Feb 06 03:01:44 <dyoung> but it... well.. It SUCKS.
Feb 06 03:05:51 <rwhitby-away> anyway, I don't have the time tonight to deal with it.
Feb 06 03:06:49 <rwhitby-away> but if someone wants to remove all .a's and .la's from packages that want to install them in /opt/lib, and bump the ipk versions for those packages, then I'll be happy to try compiling pkgconfig and gettext again on builds and prodslug tomorrow.
Feb 06 03:07:40 <rwhitby-away> until then, they stay out of the feeds (unless another solution is found).  it's unfair on those packages, but that's life.
Feb 06 03:08:47 <jacques> unfortunately, native toolchain is going to be one of them
Feb 06 03:09:03 <jacques> or is it already removing the .la's ?
Feb 06 03:09:58 <rwhitby-away> looks like it's not, so lets start there.
Feb 06 03:10:36 <rwhitby-away> . -bash-2.05b$ ls /opt/lib/*.{a,la}
Feb 06 03:10:36 <rwhitby-away> . /opt/lib/libMagick++.a   /opt/lib/libbz2.a     /opt/lib/libltdl.la    /opt/lib/libncurses_g.a  /opt/lib/libpng12.la
Feb 06 03:10:36 <rwhitby-away> . /opt/lib/libMagick++.la  /opt/lib/libfl.a      /opt/lib/libmagic.a    /opt/lib/libpanel.a      /opt/lib/libpopt.a
Feb 06 03:10:36 <rwhitby-away> . /opt/lib/libMagick.a     /opt/lib/libform.a    /opt/lib/libmagic.la   /opt/lib/libpanel_g.a    /opt/lib/libpopt.la
Feb 06 03:10:37 <rwhitby-away> . /opt/lib/libMagick.la    /opt/lib/libform_g.a  /opt/lib/libmenu.a     /opt/lib/libpng.a        /opt/lib/libtermcap.a
Feb 06 03:10:40 <rwhitby-away> . /opt/lib/libWand.a       /opt/lib/libjpeg.a    /opt/lib/libmenu_g.a   /opt/lib/libpng.la       /opt/lib/libtiff.a
Feb 06 03:10:43 <rwhitby-away> . /opt/lib/libWand.la      /opt/lib/libltdl.a    /opt/lib/libncurses.a  /opt/lib/libpng12.a      /opt/lib/libtiff.la
Feb 06 03:10:48 <rwhitby-away> There's the hit-list on prodslug for native compiles
Feb 06 03:11:32 <jacques> once again for the record, not *all* .a files are bad
Feb 06 03:12:09 <rwhitby-away> right, but ones in /opt/lib are
Feb 06 03:12:20 <rwhitby-away> . -bash-2.05b$ ls /opt/armeb/armv5b-softfloat-linux/lib/*.la
Feb 06 03:12:20 <rwhitby-away> . /opt/armeb/armv5b-softfloat-linux/lib/libstdc++.la  /opt/armeb/armv5b-softfloat-linux/lib/libsupc++.la
Feb 06 03:12:27 <jacques> those need to go
Feb 06 03:12:27 <rwhitby-away> There's the hit-list for crosstool-native.
Feb 06 03:12:39 <jacques> but for example, /opt/lib/libpcap.a
Feb 06 03:12:46 <jacques> there is no libpcap.so
Feb 06 03:13:21 <jacques> libtermcap.a - same
Feb 06 03:13:22 <rwhitby-away> right, but packages that need that can get it from staging
Feb 06 03:13:34 <jacques> oh right we did say that
Feb 06 03:14:10 <rwhitby-away> if that works, then we can add those rules to the packaging best practises page.
Feb 06 03:14:31 <rwhitby-away> night all
Feb 06 03:14:35 <jacques> `night
Feb 06 03:15:53 <jacques> hmm there is a downside to that policy tho
Feb 06 03:15:59 <rwhitby-asleep> But I still don't know how jp30 got it to build an ipk on cross-compile (cause he specifically added pkgconfig to cross-and-native ready for testing)
Feb 06 03:16:19 <jacques> what about ppl who want to do native devel not using unslung cvs ?
Feb 06 03:16:30 <jacques> for them, staging doesn't exist
Feb 06 03:16:33 <rwhitby-asleep> unfortunately, we don't support that.
Feb 06 03:16:43 <rwhitby-asleep> they're on their own.
Feb 06 03:16:59 <rwhitby-asleep> we have to draw a line somewhere.
Feb 06 03:17:15 <jacques> seems a bit harsh when all we need to do is be sure we don't move .a's from /opt/lib when there is no equivalent .so
Feb 06 03:18:08 <dyoung> we can start with .la's
Feb 06 03:18:10 <rwhitby-asleep> ok, let's go with that compromise and see how many times we get back into this situation again with new developers
Feb 06 03:18:23 <rwhitby-asleep> just blow away .la's for the moment
Feb 06 03:18:30 <dyoung> I think the la's are the big offender.
Feb 06 03:18:40 <rwhitby-asleep> nod
Feb 06 03:18:45 <dyoung> I'm not sure if I've had a serious .a show-stopping incident.
Feb 06 03:18:49 <jacques> even .la's and the huge .a's with equivalent .so's I'm fine with
Feb 06 03:18:59 <jacques> some of those static libs are huge
Feb 06 03:19:01 <rwhitby-asleep> yep, me too
Feb 06 03:19:08 <dyoung> Agreed.
Feb 06 03:19:11 <rwhitby-asleep> done
Feb 06 03:19:15 <rwhitby-asleep> night
Feb 06 03:19:21 <dyoung> me too.
Feb 06 03:19:24 <jacques> I hope nobody notices the huge libperl.a
Feb 06 03:19:26 <jacques> :-)
Feb 06 03:19:38 <jacques> since we are building perl static for performance 
Feb 06 03:19:44 <dyoung-zzzz> I'm kind of wondering how libmagick++ got it in there.
Feb 06 03:20:01 <dyoung-zzzz> I thought I specifically didnt compile that stuff in.
Feb 06 03:20:09 <jacques> 15% performance decrease when using libperl.so
Feb 06 03:20:15 <dyoung-zzzz> Wow.
Feb 06 03:20:26 <dyoung-zzzz> Thats a pretty hefty penalty.
Feb 06 03:20:34 <jacques> but I understand why gentoo builds two perl packages - one is just to get the shared libperl, and the other is to get the fast perl binary
Feb 06 03:20:59 <jacques> it's real. I remember benchmarking 
Feb 06 03:21:04 <jacques> but the 15% is in the docs
Feb 06 03:21:18 <jacques> don't jsut take my word for it :-)
Feb 06 05:29:11 <caplink811> hi all
Feb 06 05:32:05 <CIA-7> 03caplink811 * 10unslung/sources/openldap/ (postinst prerm rc.openldap): initial checkin
Feb 06 05:33:09 <dyoung-zzzz> Insomnia is Evil.
Feb 06 05:34:40 <CIA-7> 03caplink811 * 10unslung/sources/openldap/control: add depencies
Feb 06 05:37:37 <CIA-7> 03caplink811 * 10unslung/make/openldap.mk: add stages, checked ipk build
**** ENDING LOGGING AT Sun Feb 06 05:39:56 2005