--- Log opened Fri Jul 29 00:00:07 2005
00:21 -!- xhrl [~ThomasW@S0106000f3d5d5bed.vs.shawcable.net] has joined #fink
01:00 -!- xhrl [~ThomasW@S0106000f3d5d5bed.vs.shawcable.net] has quit ["leaving"]
01:00 < gzl> how does one write a system-* package?
01:01 < gzl> er, badly phrased
01:01 < gzl> are there any special policies involved in writing one?
01:22 -!- shres [~sshreyas@202.144.86.147] has joined #fink
02:38 -!- vasi [~vasi@modemcable133.147-70-69.mc.videotron.ca] has joined #fink
02:40 < vasi> hey cirdan
02:46 < vasi> !logs
02:47 < Melian> logs are at http://meme.b9.com/cview.html?channel=fink&date=today or http://fink.aquaflux.org (currently down), or http://hollowvoice.org/~ranger/irc_logs/
02:53 -!- linuxmaniac [~maniac@234.Red-83-38-147.pooles.rima-tde.net] has joined #fink
02:53 -!- linuxmaniac [~maniac@234.Red-83-38-147.pooles.rima-tde.net] has left #fink ["Leaving"]
04:24 -!- vasi [~vasi@modemcable133.147-70-69.mc.videotron.ca] has quit []
04:28 < shres> Whats the irc nick of Michele Garoche ?
04:38 -!- nukenuke [~fth@203.210.250.247] has joined #fink
04:57 -!- kane_ [~kane@perl.xs4all.nl] has joined #fink
05:31 -!- vasi [~vasi@modemcable133.147-70-69.mc.videotron.ca] has joined #fink
05:41 -!- chor [~snag@nl-67-124.netlogon.liu.se] has joined #fink
05:43 < chor> Hi, When I try to update fink I get an error saying Im using gcc 3.3 and that I should update to gcc 4. It also says I should run "gcc_select 4.0" but that doesn't work. What to do?
05:51 -!- nukenuke [~fth@203.210.250.247] has quit []
06:22 < vasi> chor, in what way does it not work?
06:22 < vasi> ie: what happens?
06:22 < vasi> lisppaste: url
06:22 < lisppaste> To use the lisppaste bot, visit http://paste.lisp.org/new/fink and enter your paste.
06:22 < chor> I get the "usage" information for gcc_select
06:23 < vasi> can you paste that output to the url above?
06:24 < chor> im trying
06:24 < chor> :D
06:25 < lisppaste> chor pasted "gcc_select" at http://paste.lisp.org/display/10305
06:25 < vasi> ah....lemme guess, you just updated to tiger?
06:25 < chor> yes
06:26 < vasi> cool, so what you have to do is install the new XCode from your Tiger DVD
06:26 < chor> ah...
06:26 < chor> that's in another city
06:26 < chor> :(
06:26 < vasi> that's XCode 2.0...it wouldn't hurt to go to connect.apple.com and get the XCode 2.1 update after that
06:26 < vasi> oooh, well that's unfortunate!
06:26 < vasi> see if you can find somebody else with the DVD?
06:26 < chor> so if I install XCode it will work?
06:27 < vasi> XCode 2.x
06:27 < vasi> 1.x (which is what you already have) won't work with Fink on Tiger properly
06:27 < chor> thanks. Than I will have to do that when I get home
06:27 < vasi> maybe apple has it available as a massive download, lemme check
06:28 < chor> ok
06:28 < vasi> but i can't seem to check right now, because my computer is convinced my net connection is down
06:28 < vasi> despite the fact that i'm talking to you :-)
06:29 < vasi> brb
06:45 < chor> I found Xcode 2.1 on apples site and have now downloaded it. I'm trying to install gcc 4.0 now
06:45 -!- vasi [~vasi@modemcable133.147-70-69.mc.videotron.ca] has quit [Read error: 104 (Connection reset by peer)]
06:51 < chor> I now have gcc 4.0 :D
06:51 < chor> Thanks
06:54 -!- hennker [flullup@dsl-213-023-248-191.arcor-ip.net] has joined #fink
07:20 < cirdan> yo
07:31 < chor> But now I get another error :(
07:32 < lisppaste> chor pasted "fink selfupdate; fink index; fink scanpackages" at http://paste.lisp.org/display/10307
07:33 < chor> I edited fink.conf to contain unstable/main unstable/crypto and run "fink selfupdate; fink index; fink scanpackages"
07:34 < chor> I get the error above it appears as if the unstable packages isn't added
07:57 -!- chor [~snag@nl-67-124.netlogon.liu.se] has quit ["Computer goes to sleep!"]
08:04 -!- vasi [~vasi@modemcable133.147-70-69.mc.videotron.ca] has joined #fink
08:28 < shres> vasi: Would u know Michele Garoche's irc nick ?
08:29 < vasi> miga
08:29 < vasi> you can always email her if you need something....
08:31 < shres> Ah, Thanks. I just needed to clarify a ORBit2 patch she submitted.
08:38 < vasi> !alicekill SF.net
08:38 < Melian> I want to kill SF.net, I mean, I wanna, I wanna kill SF.net. Kill. I wanna see SF.net's blood and gore and guts and have SF.net's veins in my teeth.  Eat SF.net's dead burnt body.  I mean kill, Kill, KILL, KILL! and I jump around with vasi, yelling "KILL, KILL"
08:45 < pogma> what did sourceforge do now?
08:48 < vasi> cvs isn't liking me
08:48 < vasi> hey pogma, thesin and cirdan and myself were discussing linking to -ssl and non-ssl shlibs
08:49 < vasi> in my experience, i've never had a problem switching from one to another....have i just been lucky?
08:50 < pogma> I think so
08:51 < pogma> Have you tried switching something from -ssl to non-ssl and then using the program to do https:// ?
08:52 < pogma> What does the program do if you try this? (if it did not hardcode "we've got -lssl" at configure time, then it probably won't crash)
08:52 < pogma> which is good :)
08:52 -!- newmanbe [~newmanbe@4d9aae06ae21aecb.session.tor] has joined #fink
08:53 -!- geewz [~gregreede@ppp122-142.static.internode.on.net] has joined #fink
08:54 < vasi> hmmm, so should -ssl and non-ssl libs have the same install-name then?
08:54 < vasi> if they actually aren't compatible...
08:55 -!- nkuttler [nicolas@server.asta.uni-saarland.de] has quit ["leaving"]
08:56 < pogma> I think we need to just ditch our openssl and always use apple's, thus resolving the license problems and at the same time we can move -ssl packages out of crypto
08:58 < pogma> bbiab
08:58 < vasi> k
09:05 -!- zizban [~zizban@24-52-0-219.sbtnvt.adelphia.net] has joined #fink
09:05  * zizban yawns as he returns from morning mass
09:18 < zizban> bbl
09:18 -!- zizban [~zizban@24-52-0-219.sbtnvt.adelphia.net] has quit [Remote closed the connection]
09:22 -!- akh [~akhansen@ldx3.psfc.mit.edu] has joined #fink
09:23 -!- zizban [~zizban@24-52-0-219.sbtnvt.adelphia.net] has joined #fink
09:26 -!- waista [~user@19.15.233.220.exetel.com.au] has joined #fink
09:28 -!- vasi [~vasi@modemcable133.147-70-69.mc.videotron.ca] has quit []
09:28 < waista> i'm having trouble with the sbcl package. When loading Slime i get "debugger invoked on a SB-INT:EXTENSION-FAILURE in thread 1222:                  |
09:28 < waista>   Don't know how to REQUIRE SB-POSIX."
09:29 < waista> Do a require from outside of emacs and slime returns the same problem
09:30 < zizban> I have no idea. I'd email the maintainer
09:31 < waista> okay thx
09:31 < zizban> np
09:31 < waista> i'll do it if you like
09:31 < newmanbe> Hmm, what does M mean for cvs update...
09:31 < zizban> sure
09:31 < waista> bye
09:31 -!- waista [~user@19.15.233.220.exetel.com.au] has left #fink ["ERC Version 5.0.1 $Revision: 1.726.2.3 $ (IRC client for Emacs)"]
09:31 < zizban> bye
09:31 < zizban> heh
09:31 < zizban> my kind of guy
09:35 < newmanbe> !cvsweb
09:36 < zizban> Melian must be asleep
09:36 < newmanbe> !unstable
09:36 < Melian> i guess unstable is http://www.finkproject.org/faq/usage-fink.php#unstable , or at http://www.finkproject.org/faq/usage-fink.php#bindist
09:37 < newmanbe> !viewcvs
09:37 < Melian> viewcvs is probably a web interface to Fink's CVS tree at http://cvs.sourceforge.net/viewcvs.py/fink
09:37 < akh> newmanbe: man cvs
09:37 < akh> If an answer is in a manpage that's $50.
09:37 < newmanbe> Nah, it has nan info page. :)
09:38 < newmanbe> an
09:38 < akh> Same principle--documentation installed on the system.
09:38 < zizban> ooo $50
09:39 < akh> If we actually had a way to fine people, we could buy a bindist server farm in a month.
09:39 < zizban> heh
09:39 < newmanbe> akh: But RangerRick always said we wouldn't be able to use it yet.
09:40 < akh> How so?
09:40 < zizban> so? we can develop nuklear weapons!
09:40 < newmanbe> I dunn'o.
09:40 < akh> We could use it for the same old bindist generation that we've been doing.
09:40 < newmanbe> He usually mentions something about not being able to build it automatically.
09:41 < newmanbe> Simmilar to SourceForge.net's Compile Farm?
09:41 < akh> Right.
09:42 < akh> Automated builds are nice, but not required--if they were we wouldn't have a bindist at all.
09:42 < zizban> I need a light weight WM for solaris...any suggestions?
09:43 < newmanbe> solariswm?
09:43 < zizban> nope..I will not use CDE :-)
09:44 < newmanbe> They still use CDE?
09:45 < zizban> ya, it's still default, though JDS is there now
09:45 < akh> b4step?
09:45 < akh> oops:  B4step
09:45 < akh> Or how about lwm ?
09:45 < newmanbe> AfterStep? :-p
09:45 < akh> dang:  xwinman.org is slow today.
09:45 < zizban> B4step?
09:46 < zizban> Oh I am there now, I just don 't want to browse through 200 listings
09:46 < zizban> maybe I'll go with olvwm
09:47 < zizban> let me check out B4step. I thought you were kidding
09:47 < akh> I wouldn't do that on a matter this serious.  ;-)
09:47 < akh> (and I wouldn't have made up something that silly)
09:48 < zizban> heh
09:48 < zizban> hmmm interesting
09:50 < akh> And hasn't changed in a year so it must be stable.  :-)
09:50 < zizban> cool
09:51 < zizban> I dont mind if it's old as long as it works
09:51 < akh> (or permanently b0rked)
09:52 < zizban> heh
09:56 < akh> ugh--they keep updating xcircuit.  If I were actually getting any feedback it would be a nightmare to try to keep migrating new versions to stable.
09:56 < zizban> heh
09:56 < zizban> I have never used xcircuit
09:57 < akh> Fair enough.
09:59 < zizban> I'm not sure I could make use of it :)
09:59 < akh> Yeah--that's the problem with these specialized packages.
10:01 < zizban> yep
10:02 < zizban> I like xfce4, I hate the top taskbar
10:03 < akh> It can't be moved?
10:04 < cirdan> !seen vasi
10:04 < Melian> vasi <~vasi@modemcable133.147-70-69.mc.videotron.ca> was last seen on IRC in channel #fink, 1h 5m 24s ago, saying: 'k'.
10:07 < zizban> no, but you can make it a plugin (I believe) in the panel
10:07 < akh> ah
10:07 < zizban> I have a small screen for my PC so it's an issue...oh well
10:16 -!- miga [~miga@easyconnect2121138-139.clients.easynet.fr] has joined #fink
10:16 -!- miga [~miga@easyconnect2121138-139.clients.easynet.fr] has quit [Client Quit]
10:19 -!- shres [~sshreyas@202.144.86.147] has left #fink ["Leaving"]
10:36 < akh> The mail and CVS servers seem to be working well today.
10:36  * akh jinxes all Fink activity for the weekend.  ;-)
10:42 -!- regeya [~shane@adsl-sp3-cdale176.micgi.com] has joined #fink
10:44 < zizban> heh
10:44 < zizban> today is friday...right. getting up early screwed me up
10:46  * akh slept in today--6AM.
10:47 < zizban> heh. I went to mass this morning 8am
10:47 < newmanbe> Oopsies, I hate CVS.
10:47 -!- kane_ [~kane@perl.xs4all.nl] has quit []
10:48 < akh> newmanbe: Are you breaking the website again?
10:48 -!- kane_ [~kane@perl.xs4all.nl] has joined #fink
10:50 < newmanbe> No I apparently imported something to a module I shouldn't have access to.
10:51 < zizban> ooops
10:51 < akh> Interesting...
10:52 < newmanbe> !lisppaste
10:52 < Melian> You can use lisppaste to paste errors at http://paste.lisp.org/new/fink , instead of flooding the channel.
10:52 < newmanbe> At least, last I checked I shouldn't be able to write to.
10:53 < lisppaste> newmanbe pasted "E-mail from Max Horn" at http://paste.lisp.org/display/10310
10:54 < newmanbe> akh: I can't edit the website at all, except for web/map .
10:54 < newmanbe> And experimental.
10:54 < akh> Right.
10:54  * akh doesn't see the module in "fink"
10:55 < akh> (rather, your additional directory)
10:56 < akh> It must have been fixed.
10:58 < akh> Or maybe there's a convoluted back door via xml/fink...
11:47 < zizban> woot! akh, wanna move to the Berkshires? They need a chemistry teacher
11:50 -!- newmanbe_ [~newmanbe@5fee916586face38.session.tor] has joined #fink
11:56 -!- mGiff [~mGiff@ottawa-hs-209-217-93-182.d-ip.magma.ca] has joined #fink
11:58 -!- mGiff [~mGiff@ottawa-hs-209-217-93-182.d-ip.magma.ca] has left #fink ["Leaving"]
11:59 -!- nothingmuch [~nothingmu@yovalk.bb.netvision.net.il] has joined #fink
11:59 < akh> zizban: tempting, but my wife wouldn't let it happen.
11:59 < zizban> plus you have tenure, right?
11:59 < akh> no
11:59 -!- drm [~drm@m015f36d0.tmodns.net] has joined #fink
12:00 < nothingmuch> i just installled fink on panther and installed gnucash, and i'm getting runtime errors:
12:00 < akh> use lisppaste
12:00 < zizban> oh
12:00 < nothingmuch> "ERROR: file: "libgw-gnc", message: dlcompat: invalid handle"
12:00 < akh> lisppaste: URL
12:00 < lisppaste> To use the lisppaste bot, visit http://paste.lisp.org/new/fink and enter your paste.
12:00 < akh> (if there's any more than that)--that looks vaguely familiar, though.
12:01 < nothingmuch> akh: that was adressed to me?
12:01 < akh> yup
12:01 < nothingmuch> ah.... that's it
12:01 < nothingmuch> except for a line preceding it saying that the error happenned in the procedure that does dynamic linking
12:02 < zizban> ttp://sourceforge.net/mailarchive/forum.php?forum_id=3200&style=flat&viewday=24&viewmonth=200409
12:02 < zizban> that's a thread about the problem from last year
12:03 < zizban> and looksee here, there is akh :)
12:03 -!- newmanbe_ [~newmanbe@5fee916586face38.session.tor] has quit [Client Quit]
12:04 -!- newmanbe_ [~newmanbe@82f07c33bb6d06a2.session.tor] has joined #fink
12:04 < nothingmuch> zizban: sounds promising... Thanks!
12:05 -!- newmanbe [~newmanbe@4d9aae06ae21aecb.session.tor] has quit [Nick collision from services.]
12:05 < akh> Here's another:  ttp://sourceforge.net/mailarchive/forum.php?forum_id=3200&style=flat&viewday=24&viewmonth=200409
12:05 < zizban> np
12:05 < akh> oops:  http://sourceforge.net/mailarchive/forum.php?forum_id=3200&style=flat&viewday=24&viewmonth=200409
12:05 < zizban> heh
12:05 < akh> wait--grrr
12:05 < akh> http://lists.gnucash.org/pipermail/gnucash-user/2005-January/012625.html
12:05 < akh> Much better.
12:05 < akh> (in that a solution is posted)
12:07 < nothingmuch> huraah! it works!
12:07 < nothingmuch> thanks guys!
12:09 < zizban> np
12:13 < akh> Now, if we only had a fix for the openjade LINKEDIT complaints...
12:14 < drm> hm?
12:19 < akh> just a sec--generating a gmane link..
12:20 < akh> http://article.gmane.org/gmane.os.apple.fink.beginners/16760
12:20 < akh> oops--that's LINKEND
12:21 < akh> My best answer has been "use the bindist"
12:21 < zizban> heh
12:21 < akh> It's the same version.
12:22 < drm> yeah... the problem may be that openjade was updated
12:24 < akh> I thought it might be something like that--and there's no real prior version to revert to.
12:25 -!- zizban [~zizban@24-52-0-219.sbtnvt.adelphia.net] has quit [Remote closed the connection]
12:27 < drm> i think the best solution may be to disable sgml documentation building for gnome-vfs, which is after all a rather old package
12:29 < akh> It's been seen for some other GNOME* packages as well--I think it's supposedly fixed in vte.
12:29 < drm> gnome1 packages, right?
12:30 < akh> I think so...
12:31  * drm would like to get rid of all gnome1 packages :/
12:31 < cirdan> !lart gnome1
12:31  * Melian moos at gnome1
12:32 < akh> Is gnucash the only thing holding us back?
12:32 < drm> Melian, you are getting creating
12:32 < Melian> drm: I think you lost me on that one
12:32 < drm> akh: its the biggie
12:32 < akh> Thought so.
12:32 < drm> there are others, though... gabber, i think
12:35 -!- kpaul [~kpaul@70.57.247.118] has joined #fink
12:37 < akh> ick
12:40 -!- nothingmuch [~nothingmu@yovalk.bb.netvision.net.il] has quit [Read error: 110 (Connection timed out)]
12:43 -!- vasi [~vasi@modemcable133.147-70-69.mc.videotron.ca] has joined #fink
12:44 < drm> good morning, vasi
12:44 < vasi> hi drm
12:44 < cirdan> hey vasi 
12:44 < drm> you guys are gonna do the release, right? or are you waiting on me?
12:44 < vasi> hi cirdan.....get my email?
12:44 < cirdan> yeah
12:44 < vasi> drm, i was just waiting a day for the mirrors to catch up
12:44 < drm> cool
12:45 < cirdan> gotta pull the old .info outta cvs and try to bootstrap
12:45 < vasi> oh, and i also want to know if we should remove 10.3 from REVISION?
12:45 < vasi> er, 10.2
12:45 < vasi> -gcc3.3
12:45 < drm> vasi: yes, we aren't releasing that one any more
12:46 < vasi> k, just the whole build-fu is not  immediately obvious to mere mortals :-)
12:46 < drm> we aren't quite releasing 10.4 yet, but i put that there to be prepared for when we do
12:46 < drm> vasi: did you read the various readme s?
12:46 < vasi> yeah, and part of the scripts themselves
12:46 < drm> i would be happy to have them improved
12:46 < drm> (by me or by you guys)
12:47 < drm> if you can articulate their shortcomings :)
12:47 < vasi> we should definitely try to use buildfink for future bindists!
12:47 < drm> vasi: yeah
12:47 < drm> vasi: one more bindist done the old way, though: the first 10.4 one
12:47 < vasi> yeah
12:48 < vasi> i just made a change to HEAD that added Info3....mostly to spark debate on -devel, since i'm not sure it's the right option
12:49 < vasi> i've learned by now that if we just talk, nothing ever gets decided and it peters out...if it's clear that without talk, something WILL happen, then everybody chimes in :-)
12:49 < drm> yup
12:50 < drm> that's why i've often posted messages like "i'm gonna do the following next week unless somebody objects"
12:51 < cirdan> please get rid of Info* and put a real InfoVersion: field in fink
12:51 < vasi> cirdan, we've discussed that
12:51 < cirdan> right, one of those things that was all talk you mentioned
12:51 < vasi> the whole point of InfoN is to deal with major changes that break the whole parsing
12:52 < cirdan> if the infovers field is parsed first, it doesn't matter
12:52 < vasi> but you CAN'T parse that first if the parser is broken
12:52 < vasi> :-)
12:52 < cirdan> yes you can
12:53 < vasi> ok, so imagine we have Info4 that contains YAML instead of our Key: Value syntax
12:53 < cirdan> grep -i infoversion: foo.fink | cut -f2 -d\:
12:53 < cirdan> and the version tells the parser what to do
12:53 < vasi> now are old versions gonna get the InfoVersion field, given that they don't know YAML from their eyebrow?
12:53 < drm> cirdan: you are objecting to the name or the syntax?
12:53 < cirdan> well, info4: << i doubt is valid YAML, so how would that help?
12:53 < cirdan> drm: both
12:54 < cirdan> there should be minor numbers in there, for some new features
12:54 < vasi> cirdan, the idea is that we use the Info4 to WRAP yaml (or whatever)
12:54 < cirdan> major #s are for when it would break the old parser
12:54 < vasi> cirdan, YES, a InfoMinorVersion makes sense
12:54 < vasi> that was my next point :-)
12:54 < vasi> however, given that InfoN+1 works just as well for that....why bother?
12:54 < cirdan> vasi: still use InfoVersion: 4 for YAML
12:55 < cirdan> vasi: logic
12:55 < vasi> look, we've been through this before
12:55 < cirdan> fink will use the YAML parser to parse the rest of the file
12:55 < drm> does the parser do the right thing with Info10, by the way?
12:55 < vasi> drm, i think so, i better check!
12:56 < vasi> yeah, uses \d+
12:56 < cirdan> right, there is no major reason *not* to go with a new field
12:56 < vasi> cirdan, huh?
12:56 < drm> cirdan: there is *always* a reason not to go with a new field
12:56 < cirdan> except the little bit of code that needs to be done
12:56 < drm> cirdan: NO NEW FIELDS
12:56 < vasi> we NEED a wrapper for major changes that break the parser
12:56 < vasi> there's no way around that
12:56 < cirdan> there is if the parser looks *only* for a certain tring before parsing the file
12:57 < vasi> inserting a random InfoVersion: N could break whatever format we use
12:57  * cirdan doesn't see what's so hard
12:57 < vasi> and also, there's no reason why a description or comment couldn't mention InfoVersion: N in a totally meaningful way, but not intending to break everything :-)
12:57 < drm> cirdan: changes like the one you are suggesting are not backward compatible
12:57 < cirdan> rell, could require it as the first line of the .info
12:58 < drm> cirdan: the advantage of the current setup is that it *is* backward compatible
12:58 < drm> an old fink can read a new-fangled file... it will just see that most fields are missing
12:58 < vasi> cirdan, so now you're suggesting infoN but without the <<'s
12:58 < vasi> is that so bad?
12:58 < vasi> adding two <<'s?
12:58 < vasi> if you're complaining about indent-ugliness, that's fixed now
12:58 < vasi> (in Info3)
12:59 < cirdan> drm: well, the soner the code is inserted, the sooner the point of backwords compat goes by
12:59 < drm> cirdan: i guess you don't believe in "stare decisis" do ya?
12:59 < vasi> heh
13:00 < cirdan> vasi: the field would allowe more precise checking of info file syntax, and use a logical version numbering system
13:00 < vasi> seriously cirdan...we have a working solution, a whole bunch of people who like it best (including dmacks, who is probably most likely to change the parser :-)
13:00 < vasi> and you want to change it cuz...you find it ugly?
13:00 < vasi> we use a logical numbering system
13:00 < cirdan> i'm not saying the switch needs to be immediate, but if the code is added, in a year it wont be much of a problem to switch things over
13:00 < vasi> sequential :-)
13:01 < cirdan> vasi: x.y :-p
13:01 < vasi> why bother? what do we gain?
13:01 < vasi> in lib-numbering, it actually MEANS something
13:01 < vasi> about the API
13:01 < vasi> for .info versioning, the difference between major and minor is zero
13:01 < cirdan> new fields that dont break the parser could use the minor version
13:01 < cirdan> if it breaks, new major
13:02 < vasi> but we already have a mechanism for new fields that don't break the parser
13:02 < cirdan> would allow uus to depreceate fields also
13:02 < vasi> that mechanism is "just add the new field"
13:02 < cirdan> and fink check foo.info would yell
13:02 -!- nothingmuch [~nothingmu@yovalk.bb.netvision.net.il] has joined #fink
13:02 < vasi> again, we can already deprecate fields in Validation
13:02 < vasi> *sigh*
13:03 < cirdan> vasi: hmm, why would anyone bother to put a standard-version of html compat in a .htlp doc then?
13:03 < cirdan> html
13:03 < vasi> because web browsers don't have selfupdate :-P
13:03 < cirdan> just don't use old fields
13:03 < cirdan> heh
13:03  * cirdan needs thesin here to even the odds
13:04 < vasi> if you really want to discuss this seriously, you have to 1) draw up a list of real things that we actually gain 2) come up with good solutions to the drawbacks 3) make sure it's obvious that the transition is worth the pain
13:04 < vasi> then you can post that to the list
13:04 < cirdan> vasi: does the parser change what's allowed if ther is a infoX field present?
13:05 < vasi> no, it just affects old parsers atm
13:05 < vasi> no reason it has to be that way though
13:05 < cirdan> i meant validation engine, sorry
13:05 -!- newmanbe [~newmanbe@6b1782d42ba9c446.session.tor] has joined #fink
13:05 < vasi> cirdan, i suspect but don't know...it should
13:08 < drm> well, thesin spoke up on fink-devel
13:08 < vasi> aw man, i keep posting from the wrong account
13:12 -!- jack- is now known as \o_O\
13:17 -!- zorton [zorton@222.muha.sndg.ls3ca31ur.dsl.att.net] has joined #fink
13:18 < vasi> oh woops, did the new selfupdate code ALSO get left out of the backporting?
13:18 -!- kpaul [~kpaul@70.57.247.118] has quit [Read error: 104 (Connection reset by peer)]
13:18 < drm> what new selfupdate code?
13:19 < vasi> in HEAD...it makes CVS only update the in-use dist
13:19 < drm> ah
13:19 < vasi> much much faster
13:20 < drm> there are going to be several issue to think through when we change distribs, by the way
13:20 < drm> including that one
13:20 < vasi> cirdan's been working on some dist-up code
13:20 < vasi> which tbh has not been looked at much
13:20 < vasi> do we have a timeframe for 10.4R?
13:21 < drm> i would like to say 1 month
13:21 < vasi> ooh soon!
13:22 < drm> (i would've liked to say 2 weeks but that is not looking realistic)
13:22 < vasi> i was kinda hoping it would be post-0.25.0, but i guess not!
13:22 < vasi> so what's the plan?
13:22 < drm> yeah, the way to get it done "soon" is to say that a bunch of GCC: 3.3 packages which don't depend on anything else, would be OK
13:23 < vasi> that's fine by me....
13:23 < vasi> will we have an upgrade plan? or will we recommend that users wipe?
13:24 < drm> no, the idea is that you switch distribs and then "fink selfupdate; fink update-all" should rebuild everything in the correct order
13:24 < vasi> ok, so two problems with that
13:24 < drm> it would be good to have the swappy code in place first, though
13:24 < vasi> 1) are we gonna bump revisions manually?
13:24 < vasi> (and depends too)
13:25 < vasi> 2) what do we do about users who 'forget' to update-all or have update-all fail, and are left with a borked installation?
13:26 < drm> vasi: (1) yes (2) nothing should be borked... things will be rebuilt the next time they are needed
13:26 < vasi> drm, if the update-all rebuilds foo, but breaks before bar (which depends on foo)
13:27 < vasi> then bar is linked to a lib with the wrong ABI
13:27 < vasi> hence borkage
13:27 < drm> but it will rebuild them in the correct order, i think, due to dependencies
13:27 < drm> oh i see
13:27 < vasi> yes, it will built them in the correct order, first foo then bar
13:27 < drm> hmmm
13:27 < vasi> but that doesn't prevent bar breaking when foo is built
13:28 < drm> well, that *shouldn't* happen
13:28 < vasi> we specifically don't prevent that, so that we can actually rebuilt without first removing the whole tree-of-packages
13:28 < vasi> drm, yes it *should*....we do it on purpose, kinda
13:28 -!- newmanbe1 [~newmanbe@eb032b0cb3a557ff.session.tor] has joined #fink
13:28 < vasi> damn i need a whiteboard or something :-)
13:29 < drm> vasi: i see the point... if the gcc-4 version of the library has incompatible symbols with the gcc-3.3 version, we are in trouble if things get interrupted in the middle
13:30 < vasi> (tbh, my personal preference is considered by most people to be insane: have a script that 1) moves the old %p out of the way 2) installs a new %p for the new dist 3) copies over conffiles and packages which don't need rebuilding 4) tries to build everything that used to be installed and isn't anymore )
13:31 < vasi> at least that's what i try to do when upgrading fink between dists :-)
13:31 < drm> well, we can try that one next time if you want to implement it for the world :)
13:31 < vasi> i'm not seriously suggesting something that drastic....but just pointing out that there is a lot of potential for breakage with our current upgrade strategy
13:33 < drm> sure...we will have to be sure to warn users about making sure that it finishes... and maybe cirdan's dist-upgrade would be useful, particularly once a bindist is in place
13:33 < vasi> well currently dist-up is just about switching the symlinks and stuff, it's just a basic framework afaik
13:33 < vasi> cirdan has some major plans ever crazier than mine, which i'm not so big on
13:34 < vasi> so we don't really have a solution, or even a real plan for a solution, just a problem :-)
13:34 < vasi> we may want to recommend that users of stable wait until the bindist is there
13:34 < vasi> and then just use apt-get dist-upgrade
13:35 < drm> the problem with distribution-switching is that the mechanism only gets people's attention when its time to switch, but that's not the right time to code a new upgrade strategy
13:35 < vasi> learn the zen way....there is no right time :-)
13:35 < drm> or: every time is the right time :)
13:36 < vasi> heh
13:36 < drm> vasi: about 0.25... what are the blockers?
13:36 < vasi> ok, 0.25 has several
13:36 -!- nothingmuch [~nothingmu@yovalk.bb.netvision.net.il] has quit [Read error: 113 (No route to host)]
13:36 < vasi> uh, dmacks and i just discussed it, so i better find the logs
13:36 < vasi> (meme.b9 is unfortunately sketchy lately)
13:37 < akh> You might be better off using Ranger*'s log--I don't know of any major gaps in it.
13:37 < vasi> no search...i'll use my local logs
13:38 < akh> ah
13:38 < akh> Lack of search _would_ be a problem.
13:39 < vasi> spotlight is sloooooow
13:39 < drm> yeeeeeeeeesssssss
13:40 < drm> especially when you have 60 GB worth of tiny documents
13:41 < vasi> lisppaste: url?
13:41 < lisppaste> To use the lisppaste bot, visit http://paste.lisp.org/new/fink and enter your paste.
13:42 < lisppaste> vasi pasted "blockers" at http://paste.lisp.org/display/10319
13:42 < vasi> btw, we're gonna deprecate 10.4T at some point not too long after 10.4R is released, right?
13:43 < drm> yes
13:43 < drm> i want the change to be quick and brutal
13:43 < vasi> yay!
13:43 < cirdan> w00t
13:43 < cirdan> ooo
13:44 < cirdan> new seede pack
13:44 < vasi> cirdan, go test the essentials stuff
13:44 < vasi> NOW! :-P
13:44 < cirdan> work atm
13:44 < cirdan> you test
13:44 < cirdan> you know where cvs is
13:44 < vasi> *sigh* check my commits, do i not look busy?
13:45 < cirdan> nopt busy enough
13:45 < drm> vasi: the list was a mixture of blockers and desired-features, right?
13:45 < vasi> well we could probably take all of those out if we had to
13:45 < cirdan> look how busy i am
13:45 < drm> i mean, i haven't kept up with what's going on with shlibs, for example
13:45 < cirdan> :-)
13:45 < cirdan> no time for commits
13:45 < vasi> it would be kinda harder to take out the shlibs and indexing
13:46 < vasi> shlibs code is still a mess
13:46 < akh> http://hollowvoice.org/~ranger/irc_logs/fink-2005-07-27.log
13:46 < vasi> i'm not sure why it wants to load/unload at the same time as the package DB rather than dynamically
13:46 < drm> ok, but does it hurt anything? or does it just need improvement for some future use
13:47 < vasi> and we're also not clear on what effect it will have on -ssl | -non-ssl stuff
13:48 < drm> releasing a partially implemented thing is ok, if it doesn't interfere with current code, and if nobody uses it yet
13:48 < vasi> yeah, that's the thing, i want TheSin to decouple it from other stuff as much as possible
13:49 < drm> what i'm saying is, this alone shouldn't stop us from releasing 0.25
13:49 < drm> we are waaaay too far from HEAD in our releases ATM
13:49 < cirdan> yeah
13:49 < drm> which is largely my fault, which is why i'm trying to encourage better behavior in the future :)
13:49 < vasi> mmm....i really want it decoupled first...it shouldn't be hard, i just don't know much about Shlibs so i can't do it myself
13:50 < vasi> i don't like the idea that loading/unloading the package DB also does some mysterious shlibs stuff that may or may not be production-ready code
13:50 < drm> well, originally thesin had a separate Shlibs.pm which copied a lot of code from other places, and then somebody (dmacks?) went through and eliminated code duplication, which of course tied it closer in to the existing codebase
13:51 < vasi> RangerRick....he made Shlibs inherit from Package, which was madness
13:52 < vasi> because there's definitely no is-a relationship
13:53 < vasi> basically, i want Shlibs to load the way that Status does....if somebody asks for it, it should load then, if it becomes invalid, somebody should say Shlibs->invalidate or something
13:53 < vasi> so that way it touches nothing if it's not needed
13:53 < vasi> as long as that's in place, i don't care HOW ugly and possibly incorrect the rest of the code is
13:53 < drm> hmm? i thought it was keeping a database of Shlibs fields from installed packages
13:54 < drm> so won't that database need to be activated, and updated every time we install/remove, and so on?
13:54 < vasi> no, only when somebody cares
13:54 < vasi> when we install/remove, we just say Shlibs->invalidate
13:54 < drm> i see
13:54 < vasi> later, when someone actually uses Shlibs->get_shlibs (the only real exported function)
13:55 < vasi> Shlibs can say "aha, it's invalid so i must update"
13:55 < drm> but there are parallels between the shlibs database and package database, right?
13:55 < vasi> so that way if 200 packages are installed, and none of them use shlibs, it doesn't scan stuff 200 times
13:55 < drm> i think that's what RangerRick was trying to exploit
13:55 < vasi> there are *some* parallels...in that they're both caches of a bunch of files
13:56 < vasi> as is Status, for that matter
13:56  * drm is unfamiliar with Status
13:56 < vasi> and VirtualPkg too
13:56 < vasi> but Shlibs is much more like Status in its usage pattern
13:57 < vasi> (ie: it's loaded and invalidated every time something is installed/removed)
13:57 < vasi> (and it's only needed for a small set of things, rather than almost all the time like Package)
13:57 < drm> ok... although i point out that eventually, Shlibs will get used a lot
13:57 < vasi> yes, it will
13:57 < vasi> but that won't suffer if it's loaded on demand either
13:57 < drm> true enough
13:58 < vasi> tbh, i intend to make Package loaded on demand as well! if not in 0.25.0, then in 0.25.x
13:58 < drm> call it 0.26
13:58 < drm> really, 0.25.x should only be bugfixes for 0.25.0
13:58 < vasi> for a five-line change? sure *shrug* gets us closer to 1.0
13:58 < drm> oh, five lines?
13:59 < drm> ok, that's a bugfix :)
13:59 < vasi> er, the initial part is five lines
13:59 < vasi> the removal of old cruft is longer
14:00 < drm> (as far as numbering goes, i thought that 0.9 was pretty close to 1.0 but then max said we should use 0.10 :)
14:00 < akh> I thought 1.0 could only be approached asymptotically...
14:00 < cirdan> lol
14:00 < vasi> 0.98.98.98
14:00 < drm> akh: there was a list of desired features for 1.0 once upon a time
14:00 < vasi> since i promised dmacks never to hit .99
14:00 < drm> vasi: 0.9975.3
14:00 < cirdan> %vers=ln(x)|1
14:01 < cirdan> +1
14:01 < cirdan> iirc
14:01 < drm> oh, didn't know about this promise
14:01 < akh> drm:  It was to make sure that HEAD > any release.
14:01 < drm> having a set of milestones leading to a 1.0 release would be a plan
14:01 < vasi> you'll notice that HEAD is now 0.24.99
14:02 < vasi> drm, well i suspect things will change a lot once we have an auto-builder and a 'testing' distro
14:02 < vasi> which i really intend to work towards :-)
14:02 < vasi> at that point, fink can be less of a package manager and more of a build tool
14:03 < vasi> since most people could use apt-get with testing to get recent packages
14:03 < drm> yeah, i noticed the numbering of HEAD... i'm going to suggest that we go back to the old numbering scheme when we get 0.25 released though
14:03 < vasi> drm, why?
14:03 < cirdan> i want HEAD to be newer. always.
14:03 < drm> well, for one thing, ChangeLog entries for releases are not making it to HEAD right now
14:04 < cirdan> i hate having a release screw up my install
14:04 < vasi> ChangeLogs for release_* are supposed to go into HEAD ChangeLogs!?
14:04 < cirdan> drm: that's prolly cause the branches are so far from HEAD
14:04 < cirdan> heh
14:05 < vasi> we have SCM for a reason....
14:05 < drm> vasi: well, under the old system, you had to change the number in both HEAD and the release branch, so of course the ChangeLog entry went both places
14:06 < drm> to the extent that the ChangeLog is history, its difficult if there is not a linear history to read to get an overview of what happened, which is fairly complete
14:06 < vasi> why is it desired to bump it in HEAD? the micro-version doesn't really reflect anything else in the HEAD changelogs
14:06 < cirdan> drm: right, the changelog reflects the code
14:07 < cirdan> but not all changes in the branches make it into EHAD
14:07 < drm> cirdan: well they should
14:07 < vasi> drm, one thing i'm trying to do is add a simpler set of notes to releases
14:07 < drm> good
14:07 < vasi> so that people can get a general idea of "what's changed" without having to know the intricacies of our code
14:08 < drm> yeah... i've tried to do a little bit of that, but without too much consistency
14:08 < vasi> buuuuut....i'm truly not convinced that it makes sense to have VERSION bumps in HEAD
14:08 < drm> ok
14:09 < drm> if all major releases are numbered 0.x.0, i agree with you... if .1, .2, .3 are bugfixes then we don't need to worry
14:09 < drm> (however, that's not been true recently...)
14:09 < vasi> drm, well we have to decide what we want the fink dev cycle to look like
14:09 < vasi> right now it's just "when it's ready"
14:10 < drm> yeah
14:10 < vasi> but we need a way to accommodate bugfixes, small backportable features, and major architectural issues
14:10 < vasi> the old model made big changes really difficult, since you had to get them all worked out in the same time as minor features
14:11 < vasi> and so minor features tend to just sit in head
14:12 < drm> right... the bottleneck in the last six months, though, has been my reluctance to let 0.24.x go to stable without implementing some kind of swappy code first
14:12  * akh thought that said "snappy" for a moment.  ;-)
14:13 < cirdan> hehe
14:13 < drm> before that blocker, we had a pretty good system: huge changes had to be tested on a branch first, ordinary changes were made on HEAD, and .1, .2, etc. releases were bugfixes for .0
14:13 < vasi> yes well that was very confusing since i cannot understand when thesin tries to talk to me without using punctuation if you get my drift its quite confusing
14:13  * drm recommends faulkner or joyce
14:13 < vasi> the branching bit was also suboptimal
14:13 < vasi> lots of branches languished
14:13 < drm> true
14:14 < vasi> CVS doesn't make branching and merging especially easy
14:14 < vasi> easier than without CVS, but still :-)
14:15 < cirdan> right, would be nice to have some magic cvs sync command
14:15 < cirdan> which tags head, and merges head to your branch
14:15 < cirdan> would be so cool
14:16 < drm> maybe someday we will switch to svn
14:16  * cirdan could come up w/something like that, actuaclly...
14:16 < cirdan> but it would not be 100% dummyproof
14:16 < vasi> cirdan, it's called arch :-)
14:16 < vasi> or baz, or all that stuff
14:16 < cirdan> [chris@dale:~$]> arch
14:16 < cirdan> ppc
14:16 < cirdan> :-)
14:17 < vasi> tla, silly
14:17 < vasi> could we do checkpoint releases maybe?
14:17 < vasi> ie: bugfixes are x.y.1, .2, .3...and then we put a bunch of features in and jump to .10 or something
14:18 < cirdan> drm: we need to do a little fundraising at least, soon-ish
14:18 < cirdan> i am putting another $100 into the acct. to cover bank charges :-/
14:18 < drm> running out of funds, are we? oops
14:18 < cirdan> then we can dump sf and buy our own equip
14:18 < cirdan> :-)
14:19 < cirdan> or at least, use sf for mirrors but have main code on our own stuff
14:19 < vasi> uh...haven't the opendarwin folks volunteered their stuff?
14:19 < drm> we can dump sf anytime we need to...we have an invite at opendarwin
14:19 < drm> dumping sf is a pain, though
14:20 < drm> means a lot of recoding... no need to do it if not necessary
14:20 < vasi> well the piece-wise way we've been doing things is ok
14:20 < cirdan> drm: yeah, but there stuff already maxes out the machines they have
14:20 -!- Albie [~ambs@bl5-163-48.dsl.telepac.pt] has joined #fink
14:20 < cirdan> well, they have a new xserve, but still
14:20 < drm> cirdan: the point is, they have a sysadmin
14:20 < cirdan> i'd rather be able to have a seperate machine next to thiers
14:21 < vasi> how about using smaller-than-micro versions? fink 0.25.0.1
14:21 < cirdan> nano?
14:21 < vasi> or just declaring the leading zero dead, and jumping to 25.0.1
14:21 < cirdan> hah
14:22 < akh> heh
14:22 < vasi> (and then after that we can have fink 2006, and fink XP)
14:22 < cirdan> we can have a version 100 release party :-)
14:22 < drm> we already have fink-mirrors-0.x.y.z
14:22 < cirdan> fink vista!
14:22 < drm> FINK VISTA!
14:22 < cirdan> hah, beat ya!
14:22 < drm> hehe
14:22 < cirdan> finkhorn?
14:22 < cirdan> longfink
14:22 < akh> And how about making the bindist version mach the darwin kernel version rather than being a factor of 10 lower?
14:23 < akh> (or for that matter, sync the two)
14:23 < akh> (fink and Fink)
14:23 < cirdan> or have fink 0.x.y and bindist 10.3
14:23 < cirdan> :-)
14:23 < cirdan> no more Fink
14:23 < drm> akh: i want to see what happens to darwin versioning with 10.6
14:23 < cirdan> drm: i bet it goes up
14:24 < cirdan> ;-)
14:24 < vasi> will there even be a 10.6?
14:24 < drm> to 10?
14:24 < akh> Good plan--there was that 1 -> 5 jump before, right?
14:24 < vasi> i'm betting on Mac OS X 11 :-)
14:24 < akh> OS X11 ?
14:24 < vasi> just cuz it's not confusing enough yet
14:24 < drm> and will you use the X11 window manager?
14:24 < cirdan> i want os 0xa1
14:24 < cirdan> drm: it will merge
14:24 < cirdan> x11windowmanager.app
14:24 < drm> not bloody likely
14:25 < cirdan> ;-)
14:25 < cirdan> i do want a wine-like layer for x11 apps
14:25 < cirdan> that would rock
14:25 < vasi> you can already make X11.app hidden
14:25 < cirdan> they would just run
14:25 < vasi> and you can double-click on x11 apps
14:25 < cirdan> vasi: but if it crashes, all x11 apps do too
14:26 < vasi> yeah, same with windowserver :-)
14:26 < cirdan> right, but windowserver is a little more tested ;-)
14:27 < vasi> drm, so how soon do you want 0.25 out? should we bypass 0.24.9?
14:28 < drm> i think we should move the swappy code to branch_0_24 in case time becomes an issue, cause i want that before the new distro, but if we can get 0.25 ready for release in a few weeks, i say we go with that
14:28 < vasi> ok, we'll see
14:28 < vasi> remember that the indexing fixes aren't super-small
14:29 < drm> ok... i am ignorant there
14:30 < vasi> well FinkCommander needs a fix
14:30 < vasi> i think i brought that up
14:30 < drm> ah, right
14:30 < vasi> and rsync still doesn't work properly
14:30 < drm> ok, in that case, we may need .9
14:31 < vasi> and there are two more small robustness fixes that NEED to go in before release
14:31 < vasi> the other changes to indexing can wait
14:33 < vasi> ooh drm, did i ask you my bringing-fink-to-newbs question yet?
14:33 < drm> nope
14:33 < vasi> well one of the things that i'm into is trying to make fink much more accessible
14:34 < vasi> and i'm leaning towards two methods: 1) automatically producing complete self-contained .apps from fink packages or 2) making a GUI for fink that's focused on programs rather than packages (ie: don't show the users -dev and -shlibs and all that, and include icons, screenshots, etc)
14:35 < vasi> i've gone more work towards #1 than #2....but what do you think would work best?
14:36 < drm> i've thought about the self-contained .app question... there are some issues
14:36 < cirdan> vasi: i tink #2 is much better
14:37 < cirdan> + easier
14:37 < drm> but i think its a good thing to try... what would you do about libs?
14:37 < vasi> drm, stash them in the .app...yeah, it's more HD usage
14:37 < cirdan> frameworks?
14:37 < cirdan> vasi: doesn't that destroy the point of shared libs?
14:37 < vasi> i have a project at http://nicotine-app.sf.net that i use to investigate these things
14:37 < vasi> cirdan, yes it does :-)
14:37 < drm> vasi: so i thought of the following: 
14:38 < cirdan> they need to use the same file, or it keeps getting loaded into memory
14:38 < drm> vasi: some kind of program which examines whether fink is installed, installs it if absent, installs all needed libs using fink
14:38 < vasi> cirdan, the idea is to make it easy for users...i'd rather have a few more MB of memory used if they can just drag-n-drop apps anywhere
14:38 < vasi> drm, yeah that's basically #2
14:38 < drm> well, no
14:38 < cirdan> vasi: depending on the app, it's a whole lot more than a few MB
14:39 < vasi> drm, well then it's kinda what i meant by #2 :-)
14:39 < drm> you would drag and drop this guy, which was named 
14:39 < cirdan> think knome/kde apps
14:39 < drm> appropriately
14:39 < drm> vasi: ok, i like #2 then :)
14:40 < cirdan> motion to do #2?
14:40 < cirdan> ;-)
14:40 < drm> but just like various self-installing things that exist elsewhere, you would drag and drop "tex.app" and then double-clicking on it would do all the setup work
14:40 < vasi> drm, well i've been thinking about both ways of doing #2...as a traditional package management GUI, or as a .app wrapper which installs everything needed and then runs
14:40 < drm> we already have a traditional package management GUI though
14:40 < drm> i mean, it could be improved on, but...
14:41 < vasi> the problem is the setup work would need to either include a whole bunch of .debs in the app....or demand a fast net connection
14:41 < drm> right
14:41 < cirdan> vasi: what about a "simple" and advanced mode of FC?
14:41 < cirdan> simple would be catagories and apps
14:41 < cirdan> with screenshots and all
14:41 < drm> vasi: do you remember the darwinports release of .dmg files?
14:41 < vasi> cirdan, yeah kinda....i'm thinking gnome-app-install type of thing
14:41 < vasi> http://niran.org/code/soc/
14:41 < drm> i mean, .pkg files
14:41 < vasi> drm, don't they still do that?
14:41 < drm> well maybe
14:42 < vasi> some kinda mountable file system they used to have i think
14:42 < drm> i don't know if the collection is up to date or not
14:42 < vasi> kinda cool, but when i tried it very borked
14:42 < vasi> maybe it's improved now
14:42 < drm> right, but the idea is there
14:42 < drm> one huge file which has everything you need
14:42 < cirdan> vasi: would this be for gui apps only?
14:42 < vasi> cirdan, probably
14:43 < cirdan> drm: have a .dmg with a script and lots of hidden .debs?
14:43 < vasi> it's kinda the same as a .app wrapper that includes all the .debs
14:43 < cirdan> open the app and if the debs are newer they get installed
14:43 < vasi> inside it
14:43 < drm> yup to both
14:43 < cirdan> that would work
14:43 < cirdan> actuaclly
14:43 < vasi> wastes a lot of space i guess
14:43 < cirdan> kell of a lot of wasted traffic on the mirrors
14:44 < drm> give people a choice of two versions to download
14:44 < cirdan> vasi: have light and fill versions
14:44 < cirdan> full
14:44 < vasi> yeah
14:44 < drm> minimal, which grabs needed debs over the net, and maximal
14:44 < cirdan> drm: but let the max one also check for updates on the net
14:44 < vasi> updating is another issue, probably include that in the wrapper
14:45 < cirdan> vasi: write a pref file
14:45 < drm> yeah, but there needs to be a user option... like itunes has
14:45 < vasi> i've been looking at synaptic so i can steal their update dialog code and cocoa-ize it
14:45 < drm> "do you want to update now?"
14:45 < vasi> unfortunately, it needs a newer dpkg
14:45 < vasi> drm, yeah like that
14:45 < cirdan> if the version is newer than the value in prefs, intall new debs
14:45 < drm> only after asking, cirdan... and with a checkbox "don't ask me again"
14:45 < cirdan> /Library/Prefs
14:46 < cirdan> drm: no
14:46 < drm> somebody who is on a modem won't want to download big debs
14:46 < cirdan> can't be
14:46 < vasi> cirdan, i think i'd just have "install now" "ask me later" "don't bother me anymore"
14:46 < vasi> like drm said :-)
14:46 < drm> vasi: yes
14:46 < cirdan> ah, i'm not talking about the net thing
14:46 < drm> cirdan: its the Apple Way
14:46 < cirdan> but if a user d/l a new max package
14:46 < cirdan> and replaces the installed one with the new one
14:46 < cirdan> debs must be upgraded
14:47 < drm> cirdan: of course... the wrapper looks to see if the included debs are more recent than the installed ones
14:47 < vasi> is there a way i could scan the complete bindist (examining all the debs)?
14:47 < cirdan> drm: don't wanna do that on every launch though
14:47 < cirdan> vasi: for what"?
14:47 < vasi> i talked to that Niran dude, and i'm gonna take his database of debian apps <-> programs
14:47 < drm> vasi: its online
14:48 < drm> lives on sancho
14:48 < vasi> and i want to use the contents of our .debs to make a debian <-> fink mapping, so i can just totally steal his work
14:48 < cirdan> vasi: niran? use the debian project's scripts
14:48 < cirdan> some perl 
14:49 < vasi> cirdan, i want to correlate fink packages with what a user would consider "a program"
14:49 < cirdan> oh
14:49 < vasi> debian does not have a script for that, so far as i know
14:49 < vasi> but Niran does
14:49 < cirdan> ah
14:49 < cirdan> no, debian has a script to make a db of the packages
14:49 < cirdan> and cgi's to search 'em
14:49 < drm> vasi: maybe a new binary field in the info file "Standalone:"
14:49 < vasi> and he's also gathered all this metadata
14:50 < cirdan> packages.debian.org ;-)
14:50 < drm> vasi: that way, maintainers can decide the issue
14:50  * cirdan will bbl
14:50 < vasi> drm, i'd probably make a new tree outside of fink tbh
14:50 < drm> of course, we can add the tag to a bunch of things to get started
14:50 < drm> vasi: ok, but then there is a maintenance issue
14:50 < vasi> because it's hard to fit the associated files (screenshots and such) in a fink tree
14:51 < drm> screenshots?
14:51 < vasi> well there has to be some website where the user can find all these apps....and it sure as hell better look nice!
14:51 < vasi> also icons
14:51 < drm> true
14:52 < drm> i see, a kind of overlay on top of fink
14:52 < vasi> one problem is that a maximal package would kinda need to include all of fink, no?
14:52 < drm> makes sense
14:52 < vasi> drm, one thing that i'd REALLY like before doing this is a user-mode fink
14:52 < drm> all of fink, but not necessarily all package descriptions
14:52 < vasi> which is one of my medium-to-long-term goals
14:53 < drm> if somebody wants to activate fink for use by itself, they're gonna have to do "fink selfupdate"
14:53 < drm> the package db is too big to include 
14:53 < vasi> er, i don't know if i'd even include ANY .infos
14:53 < vasi> use only apt
14:54 < vasi> but still, just fink + dpkg + apt plus all the dependencies isn't small
14:54 < drm> right
14:54 < drm> these maximal guys are not gonna be small
14:54 < vasi> especially because of curses...why is that essential? for dselect?
14:54 < drm> yes
14:55 < vasi> what else would need work...hmm
14:55 < drm> and right now we have two curses in essential... but that should change eventually
14:55 < vasi> we'd need to make packages not require interaction, use debconf for that i guess?
14:55 < vasi> eww they're *both* essential?
14:55 < drm> yes ATM
14:56 < vasi> why?
14:56 < drm> cause there are no dependency tags for the old shlibs
14:56 < drm> since it was essential
14:56 < vasi> oh, cuz stuff still implicitly depends on the old one?
14:56 < vasi> yeah
14:56 < drm> so something could break
14:56 < vasi> we should repair that in 10.4R
14:56 < drm> we should repair that, but not hold 10.4R hostage to it
14:56 < vasi> i know :-)
14:58 < vasi> well i'll try to proof-of-concept this
14:59 < vasi> the one big problem compared to the self-contained stuff is really the size....with nicotine, i could eliminate fink's python, and a whole bunch of other stuff....but this way, with all the debs inside, that's just HUGE
14:59 < drm> well, the minimal package will be pretty small
15:05 -!- drm [~drm@m015f36d0.tmodns.net] has quit ["Leaving"]
15:10 -!- Albie [~ambs@bl5-163-48.dsl.telepac.pt] has quit ["This computer has gone to sleep"]
15:26 -!- Albie [~ambs@bl5-163-48.dsl.telepac.pt] has joined #fink
15:42 -!- nothingmuch [~nothingmu@yovalk.bb.netvision.net.il] has joined #fink
15:50 -!- vasi [~vasi@modemcable133.147-70-69.mc.videotron.ca] has quit []
15:50 -!- dmacks [~dmacks@netspace.org] has joined #fink
15:52 < dmacks> akh: If you have a moment, could you try vte rev 5 on tiger with and without fontconfig2-dev installed? I'm especially interested in any compile-time warnings that mention any FC_* symbols (as well as overall failure for other reasons, obviously:)
15:53  * dmacks tried to stop it from using fink's fc2 completely, hope nothing's's leaking in, or wacky fc1/fc2 compat problems.
15:55 < akh> dmacks:  I'll give it a try
15:59 < akh> I'll even use both of my boxen
16:01 < akh> mmm...new seed note with many "areas of testing"
16:04 < akh> (which I will not discuss here due to the NDA)  ;-)
16:08 < dmacks> At least they can now list a few specific things, instead of "everything, and even some other things than that."
16:08 < akh> Yup.
16:09 < akh> Guess I'm going to have to bail before the builds are done but I'm capturing logs in case of "incidents".
16:10 < dmacks> Cool, thx (no hurry...someone'll bitch on -beginners if it doeesn't build:)
16:10  * akh begins the insidious rebranding of failure modes in fink.
16:10 < akh> compile failures -> incidents
16:11 < akh> This seems to be the only GNOME1ish package that doesn't fsck  up when generating docs via sgml.
16:12 < dmacks> incidents -> events; events -> happy-hours
16:12 < akh> OK--I was wrong--it's done before I'm ready to go.
16:12 < akh> No errors with or without fontconfig2-dev installed.
16:13 < dmacks> The new gnome-vfs2 claims to fix doc generation; it's BuildDepends:fink-0.24.8 though.
16:13 < dmacks> (there's an explicit ChanageLog note in vte about doc generation also)
16:13 < akh> Ah--I'll just keep telling people to use the bindist until 0.24.x is in stable.
16:14 < akh> For the life of me, I don't understand why people seem to _want_ to build the same versions of packages as are present in the bindist.
16:15 < akh> Must not have UseBinaryDist set, I guess.
16:15  * dmacks thinks it's package-by-package, not something that'll be fixed for every pkg once we get new gnome-vfs2 .But I have no idea really.
16:15 < akh> hmm...it's mostly GNOME1 stuff, isn't it?
16:18 < akh> Well, vte looks safe until the users find another build failure mode.
16:19  * akh heads for the train
16:19 -!- akh is now known as akh_gone
16:19 < dmacks> thanks
16:20 -!- \o_O\ is now known as jack-
16:23 -!- Albie [~ambs@bl5-163-48.dsl.telepac.pt] has quit ["Leaving"]
16:24 -!- chor [~snag@h234n1fls13o978.telia.com] has joined #fink
16:25 -!- newmanbe [~newmanbe@6b1782d42ba9c446.session.tor] has quit [Nick collision from services.]
16:27 -!- newmanbe1 is now known as newmanbe
16:27 -!- newmanbe_ [~newmanbe@82f07c33bb6d06a2.session.tor] has quit [Ping timeout: 14400 seconds]
16:30 < chor> Hi. I have a strange error when trying to run "fink selfupdate". Does anyone have the time to help me?
16:30 < newmanbe> !lisppaste
16:30 < Melian> You can use lisppaste to paste errors at http://paste.lisp.org/new/fink , instead of flooding the channel.
16:30 < newmanbe> Can't help you if we don't know what the error is. :)
16:32 < lisppaste> chor pasted "fink selfupdate; fink index; fink scanpackages" at http://paste.lisp.org/display/10326
16:33 < chor> If I try to install fluxbox I get a similar error about xfree
16:34 < newmanbe> And you have XCode installed?
16:35 < chor> I downloaded Xcode 2.1 from apple today
16:35 < newmanbe> And installed it?
16:35 < chor> I'm not at home so I don't have the real dvd
16:35 < chor> but I didn't install the entire Xcode, just the parts with gcc 4. I had some errors about gcc 3.3 before...
16:36 < newmanbe> I don't know.
16:36 < newmanbe> I have to go; perhaps someone else can help you.
16:36 < chor> ok, thanks anyway
16:36 < newmanbe> Or if that fails, try one of the mailing lists.
16:36 < newmanbe> !mailinglists
16:36 < Melian> Visit http://www.finkproject.org/lists/ to learn about Fink's mailing lists, including web, NNTP, and blog access to their archives.
16:36 < chor> thanks
16:42 -!- nothingmuch [~nothingmu@yovalk.bb.netvision.net.il] has quit ["leaving"]
16:46 < dmacks> chor: You failed to install all the compiler packages (from the XCode DVD) that are needed.
16:46 < dmacks> (or your install failed...Apple's Installer.app is known to omit files)
16:46 < chor> ok?
16:46 < chor> what files do I need to install?
16:46 < chor> I can make a reinstall right now.
16:48 < dmacks> gcc3.3 and gcc4.whatever; DevSDK
16:48 < chor> Ok, will try to make a reinstall
16:49 < dmacks> If you "See `config.log' for more details.", there might be a more explicit error message (the exact file that's missing, for example) that will tell you (or someone else) what's really wrong here.
16:50 < chor> I'm making a reinstall right now. I also removed all the fink files
16:50 < chor> To get a clean install
16:50 < chor> should I do anything more then "rm -rf /sw"?
16:52 < dmacks> That'll take care of it in almost all cases for a relatively new fink install.
16:55 < chor> I get an error
16:56 < chor> can add envirements variables or something, check .profile and .login, I think it sayd
16:57 < chor> where does fink keep its logs?
17:00 < dmacks> If you previously installed fink using the binary installer, your .profile (or whatever) file has probably already been upgraded for you.
17:00 < dmacks> Installer.app is provides very sparse diagnostics:(
17:01 -!- regeya [~shane@adsl-sp3-cdale176.micgi.com] has quit [Client Quit]
17:01 < chor> So it's nothing to worry about? I'm runing "fink selfupdate" now and it seems to be working :D
17:03 < dmacks> Yeah. As long as your terminal or xterm prompt recognizes commands like "fink", your shell startup file is correct.
17:18 -!- kpaul [~kpaul@70.57.247.118] has joined #fink
17:30 -!- newmanbe_ [~newmanbe@b0f27192bad09c46.session.tor] has joined #fink
17:32 -!- newmanbe [~newmanbe@eb032b0cb3a557ff.session.tor] has quit [Nick collision from services.]
17:32 -!- newmanbe_ is now known as newmanbe
18:02 -!- kane_ [~kane@perl.xs4all.nl] has quit []
18:09 -!- dmacks [~dmacks@dmacks.active.supporter.pdpc] has quit ["leaving"]
18:16 -!- kpaul [~kpaul@70.57.247.118] has left #fink []
18:42 < chor> Im sorry to disterb you again. But I have one last problem. Everything now almost work. But when starting x "startx" I only get into the rootless mode. I can also only start x when I'm root.
18:42 -!- akh [~akhansen@68-118-244-23.dhcp.oxfr.ma.charter.com] has joined #fink
18:44 -!- zizban [~zizban@24-52-0-219.sbtnvt.adelphia.net] has joined #fink
18:50 -!- emp [~emp@70.57.239.37] has joined #fink
18:53 < akh> hmm...guess I'd best clone my Panther setup so's I can try the 10.3 seed.
18:53 < akh> wait:  That's Tiger and 10.4.3
18:54 < cirdan> heh
18:54 < cirdan> would be a little mistake :-)
18:54 < zizban> okay that makes sense now
18:55 < akh> Clearly I need food and a beer.
18:55 < zizban> not in that order ;)
19:01 -!- akh [~akhansen@68-118-244-23.dhcp.oxfr.ma.charter.com] has quit []
19:01 -!- akh_debian [~akhansen@68-118-244-23.dhcp.oxfr.ma.charter.com] has joined #fink
19:01 -!- akh_debian is now known as akh
19:02 -!- akh [~akhansen@68-118-244-23.dhcp.oxfr.ma.charter.com] has quit [Remote closed the connection]
19:23 -!- akh_gone [~akhansen@ldx3.psfc.mit.edu] has quit []
19:24 -!- akh [~akhansen@68-118-244-23.dhcp.oxfr.ma.charter.com] has joined #fink
19:28 < cirdan> hey
19:28 < cirdan> ok, we need to port hula, quick
19:28 < cirdan> beat everyone to it
19:28 < cirdan> inc. apple ;-)
19:30 < akh> Sounds good.
19:31 < cirdan> hmm, i thought it was more of a webmail app
19:31 < cirdan> but it seems to be an opensource version of groupwise
19:31 < cirdan> kinda
19:31 < cirdan> http://www.hula-project.org/General_FAQ
19:32 < zizban> I've heard of Hula before
19:45 -!- hennker [flullup@dsl-213-023-248-191.arcor-ip.net] has quit [Read error: 110 (Connection timed out)]
19:54 < zizban> !lart .cue files
19:54  * Melian takes out .cue files with the trash
20:09 < cirdan> ?
20:09 < cirdan> cdrdao can burn them i think
20:09 < cirdan> cdrecord too
20:10 < zizban> I am using Firestarter, which is a front end to both
20:11 < zizban> I am working on "getting" toast
20:13 -!- uncon [uncon@corp.efnet.net] has quit [Remote closed the connection]
20:24 -!- geewz [~gregreede@ppp122-142.static.internode.on.net] has quit ["Leaving"]
21:12 -!- akh [~akhansen@68-118-244-23.dhcp.oxfr.ma.charter.com] has quit ["Leaving"]
22:11 -!- vasi [~vasi@modemcable133.147-70-69.mc.videotron.ca] has joined #fink
22:13 -!- zizban [~zizban@24-52-0-219.sbtnvt.adelphia.net] has quit [Remote closed the connection]
22:26 -!- broken_chaos [~Chaos@S010600045ae0d580.ok.shawcable.net] has joined #fink
22:58 -!- dk0rr [~dk0r@cpe-24-194-171-176.nycap.res.rr.com] has joined #fink
22:58 -!- dk0rr is now known as rekoton
23:05 -!- rekoton [~dk0r@cpe-24-194-171-176.nycap.res.rr.com] has quit []
23:56 -!- dk0rr [~dk0r@cpe-24-194-171-176.nycap.res.rr.com] has joined #fink
--- Log closed Sat Jul 30 00:00:17 2005