T2 IRC Log: 2004-10-17

This is the log as captured by an IRC bot in the channel. The statements are those of the individual people and might not neccessarily reflect the policy and legal rules as set forth by the T2 SDE Project.

« prev | next »

--- Log opened Sun Oct 17 00:00:10 2004
00:02 -!- sparc-kly [~sparc-kly@northcarolina.hardlink.com] has joined #t2
00:02 < rxr> hm - interesting
00:03 < rxr> ld.so.cache is right
00:03 < rxr> must be the linker that does s.th. on it's own ...
00:03 * rxr investigating
00:03 < jsaw> :) okay. I gotta go to the lab. cu l8r
00:03 < rxr> it's sunday, isn't it?
00:03 < jsaw> yes.
00:03 < rxr> ok - 0:03 .. ;-)
00:04 < rxr> ok - have fun
00:04 < jsaw> but I'm home saturdays.
00:04 < jsaw> (but actually I should work 6 days to finish...)
00:04 < jsaw> bye.
00:04 < rxr> cu
00:40 < rxr> hm
00:43 < rxr> hm
00:43 < daja77> yes?
00:44 * rxr wondering about the gnu linker
00:44 < rxr> at least the lastest reincarnation of it ...
00:46 * daja77 thinking of packaging a cool game, even when it is coded in python
00:46 < rxr> urgs
00:46 < rxr> what game is it?
00:50 < rxr> urghs - this "nice" emultempl stuff in binutis does not make me feel good ...
00:50 < daja77> http://pathological.sourceforge.net/
00:51 < daja77> my repo grows anyway :)
00:54 < rxr> you still believe in rock?
00:55 < CIA-9> rene * r4388 /trunk/package/kde/kdebase/kdebase.conf: * fixed kdebase.conf quoting - sorry messed it up yesterday :-(
00:57 < daja77> is this a matter of believe?
00:58 < rxr> to some degree ;-)
00:59 < daja77> well now that i've took over the kde updates, it would be no good time to leave :)
01:00 < daja77> but i am eager to see what t2 is achieving
01:01 < daja77> n8
01:01 < rxr> rock has two kde updates in submaster ;-)
01:01 < rxr> n8 daja77 - sleep well
01:03 < rxr> oh my god
01:03 < daja77> oh ic
01:03 < rxr> the linker writes the error message out char by char:
01:03 < rxr> 22163 write(2, ",", 1) = 1
01:03 < rxr> 22163 write(2, " ", 1) = 1
01:03 < rxr> 22163 write(2, "n", 1) = 1
01:03 < rxr> 22163 write(2, "e", 1) = 1
01:03 < rxr> 22163 write(2, "e", 1) = 1
01:03 < rxr> 22163 write(2, "d", 1) = 1
01:03 < rxr> 22163 write(2, "e", 1) = 1
01:03 < rxr> 22163 write(2, "d", 1) = 1
01:03 < rxr> I should stop looking at that problem ...
01:04 < jsaw> ? why is that so strange ?
01:04 < jsaw> (re)
01:04 < rxr> I would expect fillinf os some buffer and then writting out one big chunk ...
01:04 < rxr> but I do not care - I search for the other prblem and just noticed it in the log ...
01:05 < jsaw> so, you should not look at dietlibc too closely :)
01:05 < jsaw> s/,//
01:05 < rxr> I looked so some area - and they looked quite well ...
01:05 < rxr> at least they did not caused so much grummeling in my stomarch as glibc or binutils do ...
01:05 < jsaw> the diet program itself does e.g.
01:05 < jsaw> hehe
01:06 < rxr> hm - if I should try downdating the glibc in a testbuild ...
01:06 < rxr> maybe the ld.so.cache layout changed and the ld does not like that ...
01:07 < jsaw> I actually think, ld does not look at all at this.
01:07 < rxr> it at least open it ..
01:08 < jsaw> Otherwise it would not make any sense to give -L options to it.
01:08 < jsaw> in the code, or by ld.so
01:08 < jsaw> ?
01:08 < rxr> ld does open it ...
01:08 < rxr> both ld.so.conf and ld.so.cache ...
01:09 < jsaw> *g*
01:11 < rxr> when was your last rock build including bigger desktop stuff?
01:11 < jsaw> rock or t2?
01:11 < rxr> rock
01:11 < jsaw> rock, 2 months ago.
01:12 < rxr> including stuff like gnome 2.6 and avifile?
01:12 < jsaw> yes, but not tested. my PC at home broke down in the mean-time, and my laptop can't take stuff like gnome or kde...
01:13 < jsaw> (PII 266+64Mb)
01:13 < rxr> hm
01:13 < rxr> at least it linked I guess ..
01:15 < rxr> I deflate an old binary .gem into the build - and copare what changes ...
01:16 < rxr> ah - shit - damn - it is a) old and b) built w/ gcc-3.2.2 ...
01:17 < rxr> ok - me merging back to r3772 in glibc and build from source ..
01:17 < rxr> in until glibc finished I can start a new book ...
01:19 < jsaw> ?
01:20 < rxr> ? because of the stray "in" or because of more?
01:20 < rxr> == 01:21:41 =[5]=> Building base/glibc [2.3.4-2004-05-01 2.1.0-beta].
01:20 < jsaw> becaus of merging back glibc
01:20 < rxr> I just want to find out what change caused this new behaviour - there is so much noise in all the updates that I first want to make sure where the regression was caused ..
01:21 < jsaw> this is old.
01:21 < jsaw> at least if both are the same errors (the gnome thing and this)
01:21 < rxr> what is old, the link error?
01:22 < jsaw> the link errors in gnome are old. yes.
01:22 < rxr> they did never happen - and the gnome people also saied it works for them ..
01:22 < rxr> my avifile is also s.th. that should just work. ...
01:24 < jsaw> maybe you update to arts-1.3.1 first?
01:25 < rxr> ouhm
01:26 < rxr> thanks - me doing the update on the way ...
01:28 < rxr> # diff -urN arts-1.3.{0,1} | diffstat
01:28 < rxr> CVS/Entries | 6 +++---
01:28 < rxr> arts.lsm | 4 ++--
01:28 < rxr> configure | 2 +-
01:28 < rxr> configure.in | 2 +-
01:28 < rxr> configure.in.in | 2 +-
01:28 < rxr> 5 files changed, 8 insertions(+), 8 deletions(-)
01:28 < rxr> ^- version changes only - not a single code line changed ;-)
01:28 < jsaw> ts
01:28 < rxr> arts is dead - kde 4.0 will skip it kde people told me ...
01:29 < jsaw> maybe using an even newer glibc would make sense: http://sources.redhat.com/cgi-bin/cvsweb.cgi/libc/NEWS?rev=1.147&content-type=text/x-cvsweb-markup&cvsroot=glibc
01:29 < rxr> I tried a newer - it failed in some ugly way on my sparc64 box - generation some code or doc or so ..
01:29 < rxr> at least it was too ugly for a quick fix ...
01:31 < rxr> do you see s.th. particularly interesting in that file?
01:32 < rxr> why the hell - do the fedora people are allowed to tag their stuff in the glibc cvs ...
01:33 < jsaw> not that I need any of them, but some seem to be quite good (to me, well). e.g. the nothrow item.
01:34 < jsaw> or nscd memory sharing.
01:34 < rxr> hm - there does not seem to be a 2.3.4 tag yet ....
01:34 < jsaw> that news file is only update if implementation is there
01:34 < jsaw> afaics
01:36 < rxr> hm my svn commit is stalled - /me wonders about the cause ...
01:36 < jsaw> ran out of entropy?
01:37 < jsaw> Here it stalls easily if I do not have my hardware rng running...
01:42 < rxr> FIN_WAIT2
01:42 < rxr> hm
01:42 < jsaw> oups
01:44 < CIA-9> rene * r4389 /trunk/package/kde/arts/arts.desc:
01:44 < CIA-9> * updated arts (1.3.0 -> 1.3.1) - oops forgot it in the kde updates,
01:44 < CIA-9> it is a version change only (e.g. no code) anyways - untested ;-)
01:55 < rxr> == 10/18/04 01:56:25 =[5]=> Finished building package glibc.
01:57 < rxr> hm - interesting - downdating glibc and binutils does not seem to cure it ...
02:06 < sparc-kly> t2 dont have isos for sparc64?
02:12 < jsaw> not that I know of. rxr?
02:20 < rxr> not yet
02:20 < rxr> the only rock or t2 bootable iso is the test one sparc-kly already downloaded ...
02:20 < rxr> I have halfly built t2 sparc and sparc64 systems on my U30 - but it needs polishing before I can hand it to other people ...
02:21 < rxr> ok - so it was not the glibc or our globbing update ...
02:21 < rxr> still fails ...
02:21 < rxr> well - tracking that will be fun ...
02:22 < rxr> I guess I take a nap first
02:25 < jsaw> ping daja77
02:27 < rxr> jsaw: thanks for the naming comments
02:28 < jsaw> Well, I not so sure about it. Just what my feeling says.
02:32 < rxr> I tag alpha3 anyway before I crawl into ned ...
02:34 < CIA-9> rene * r4390 /trunk/package/ (print/ printing/): * renamved the print repository to printing
02:36 < CIA-9> rene * r4391 /tags/2.1.0-alpha3/: * tagged 2.1.0-alpha3
02:37 * rxr starting a fresh build with slightly downdated binutils and glibc - let's see if this shows any difference ..
02:37 * rxr off into bed ...
02:37 < jsaw> sleep well.
02:40 -!- rxr changed the topic of #t2 to: T2 http://www.exactcode.de/oss/t2 - 2.1.0-beta in ~6 days | happy testing ;-)
02:41 -!- rxr changed the topic of #t2 to: T2 | http://www.exactcode.de/oss/t2 | 2.1.0-beta in ~6 days | happy testing ;-)
02:41 < rxr> so n8 all ;-)
02:50 < sparc-kly> http://boxter.mubex.net:8080/pics/sparc/Sunarsenal.JPG <-- my sun arsenal LOL
03:06 -!- junix_ [~junix@pD9E75E2D.dip.t-dialin.net] has joined #t2
03:25 -!- freed [~junix@pD9E76402.dip.t-dialin.net] has quit [Read error: 110 (Connection timed out)]
03:58 -!- blindcod1r [~blindcode@dsl-082-082-095-211.arcor-ip.net] has joined #t2
04:20 -!- blindcoder [~blindcode@dsl-213-023-152-031.arcor-ip.net] has quit [Read error: 110 (Connection timed out)]
07:50 -!- mnemoc [~amery@200.75.27.2] has joined #t2
08:03 -!- mnemoc_ [~amery@200.75.27.26] has quit [Read error: 238 (Connection timed out)]
08:06 -!- BoS [~BoS@dialin-212-144-087-199.arcor-ip.net] has quit [Read error: 113 (No route to host)]
08:19 -!- BoS [~BoS@dialin-145-254-077-077.arcor-ip.net] has joined #t2
08:22 -!- blindcod1r is now known as blindcoder
08:22 < blindcoder> moin
08:40 < martin___> moin
08:42 < blindcoder> moin moin
08:42 < martin___> moin blindcoder
09:39 < CIA-9> jsaw * r4392 /trunk/package/graphic/libtiff/libtiff.desc: * correct libtiff homepage
09:41 < CIA-9> jsaw * r4393 /trunk/package/develop/stlport/ ( stlport.conf stlport.desc thanks-debian.patch): * add package stlport (4.6.2) (untested)
09:42 < rxr> moin
09:42 < rxr> weee - today all my new scanners are scheduled to arrive **freu*
09:42 < blindcoder> moin rxr
09:43 < blindcoder> how many have you ordered o_O
09:43 < rxr> none ordered ;-) all for driver development I already do since 1997 or so ...
09:44 < rxr> I asked for one new - and they offered to send three different ones ...
09:44 < blindcoder> nice :)
09:44 < rxr> ttp://www.exactcode.de/oss/avision/
09:45 < rxr> because they have a huge possible order when they have functional macosx drivers for special ADF models .. ;-(
09:45 < rxr> ;-) even!
09:46 < rxr> and the SANE (and so my) stuff does run under any OS ;-)
09:46 < blindcoder> nice
09:46 < rxr> I even got payed by a bank in Venezuela to improve OS/2 support 1 and a half year ago *g*
09:47 < blindcoder> heh :)
09:48 < rxr> and when I have to do adaptions or a niceer GUI for osx I get a contract for the driver when this possible big order gets real on their side ,)
09:49 < blindcoder> nice
09:49 < rxr> 959 builds total, 293 completed fine, 0 with errors.
09:50 -!- rxr changed the topic of #t2 to: T2 | http://www.exactcode.de/oss/t2 | 2.1.0-beta in ~5 days | happy testing ;-)
09:59 < CIA-9> jsaw * r4394 /trunk/package/base/00-dirtree/data.txt:
09:59 < CIA-9> * also create the rc0.d and rc6.d directories in 00-dirtree
09:59 < CIA-9> (vmware installer explodes without these even if empty)
10:00 < rxr> you do not want to patch the vmware installer? *g*
10:00 -!- martin___ [~martin@brln-d9ba02b6.pool.mediaWays.net] has quit [Read error: 110 (Connection timed out)]
10:12 < CIA-9> jsaw * r4395 /trunk/package/office/ooo/ ( ooo.conf ooo.desc):
10:12 < CIA-9> * add package ooo (OpenOffice.org 1.1.3)
10:12 < CIA-9> WARNING: incomplete configuration file -> UNTESTED
11:00 -!- rxr_ [~rene@p213.54.250.129.tisdip.tiscali.de] has joined #t2
11:00 -!- Topic for #t2: T2 | http://www.exactcode.de/oss/t2 | 2.1.0-beta in ~5 days | happy testing ;-)
11:00 -!- Topic set by rxr [] [Mon Oct 18 09:51:49 2004]
11:00 [Users #t2]
11:00 [ blindcoder] [ CIA-9 ] [ jsaw ] [ mnemoc ] [ rxr ] [ sparc-kly]
11:00 [ BoS ] [ daja77] [ junix_] [ Ragnar0k] [ rxr_] [ valentin ]
11:00 -!- Irssi: #t2: Total of 12 nicks [0 ops, 0 halfops, 0 voices, 12 normal]
11:00 -!- Channel #t2 created Sun Aug 8 21:15:33 2004
11:01 -!- rxr [~rene@p213.54.236.120.tisdip.tiscali.de] has quit [Nick collision from services.]
11:01 -!- Irssi: Join to #t2 was synced in 18 secs
11:01 -!- You're now known as rxr
11:12 * rxr converting more ugly 3rd party kernel driver packages to postlinux ...
11:17 < daja77> jsaw: ?
11:19 < rxr> ePost schließt die Pforten
11:20 < rxr> ohhhhhhhhhhhhh - but my life long mail address they promissed *grin*
11:20 < rxr> ah yes - companies define "life-long" as life of the company^Wproduct only ... ;-)
11:21 < daja77> hehe
11:21 < blindcoder> what a great thing I never had one of those addresses :)
11:22 < rxr> hehe - me of course neither ;-)
11:23 < rxr> I just wanted to preserve rene.rebe@gmail.com just in case - not because I need email addresses ... ;-)
11:23 < rxr> (but it sucks slighly that c. erased the rene@eocklinux.org alias that quickly ...)
11:24 < blindcoder> should I be worried having 107 @scavenger... addresses?
11:25 -!- mnemoc_ [~amery@200.75.27.2] has joined #t2
11:25 < rxr> hehe - I also have cool `md5sum`@{exactcode,rocklinux-consulting}.de addresses, too
11:25 < blindcoder> nah, I usally just have `url I had to register at`@scavenger
11:26 < blindcoder> so, slashdot@ symlink@ palmdev@ and so on
11:26 < blindcoder> so far happypenguin@ delivers the most spam
11:27 < blindcoder> ~3 per week :-)
11:29 < CIA-9> rene * r4396 /trunk/package/x11/ (2 files in 2 dirs):
11:29 < CIA-9> * fixed gaton-km and gatos-drm-kernel to use postlinux
11:29 < CIA-9> * disabled gatos-drm-kernel by default - it did install nothing before
11:29 < CIA-9> (big *grin* about submaster code quality filtering ...) and, if done
11:29 < CIA-9> right, would overwrite the X.org, up-to-date and working modules ...
11:29 -!- mnemoc [~amery@200.75.27.2] has quit [Read error: 110 (Connection timed out)]
11:29 < rxr> 959 builds total, 315 completed fine, 0 with errors.
11:31 -!- rxr changed the topic of #t2 to: T2 | http://www.exactcode.de/oss/t2 | lim (t -> 2.1.0) $pkg = 0
11:35 -!- martin___ [~martin@brln-d9ba023c.pool.mediaWays.net] has joined #t2
11:35 < CIA-9> rene * r4397 /trunk/package/x11/gatos-ati-remote/gatos-ati-remote.conf: * converted gatos-ati-remote to postlinux - untested
11:35 < rxr> h martin___
11:37 < CIA-9> rene * r4398 /trunk/package/x11/ (2 files in 2 dirs): * some more gaton-ati*.desc indenting ...
11:38 < blindcoder> so, what does that topic mean for the mathematically challenged?
11:39 < rxr> although due to the one dimensional nature of the topic it might not be fully corret written down ...
11:40 < blindcoder> well, I wouldn't understand it in correct way, either :-)
11:41 < rxr> it should explain that the limit of the error count should convergate towards zero
11:41 < blindcoder> ah
11:41 < blindcoder> so in a full build there should be (almost) 0 packages failing?
11:41 < rxr> converge even
11:42 < rxr> should - hehe ;-)
11:42 < rxr> 11:31 -!- rxr changed the topic of #t2 to: T2 | http://www.exactcode.de/oss/t2
11:42 < rxr> | lim (t -> 2.1.0) $pkg = 0
11:42 < rxr> 11:35 -!- martin___ [~martin@brln-d9ba023c.pool.mediaWays.net] has joined #t2
11:42 -!- rxr changed the topic of #t2 to:
11:42 < rxr> oops .. - damn
11:43 < blindcoder> Well, I also have 0 failing packages in LVP if I don't build for Pentium 4
11:43 -!- rxr changed the topic of #t2 to: T2 | http://www.exactcode.de/oss/t2 | lim (t -> 2.1.0) err (t) = $pkgs = 0
11:43 < rxr> maybe this is more clear ...
11:43 < rxr> blindcoder: lvp using t2 as base?
11:44 < blindcoder> rxr: nah, using ROCK as base. Haven't built any LVP since 0.4.2 was released
11:44 < rxr> well - sure, my rock tags for quite fat deskto builds also had nerly 0 errors ...
11:44 < blindcoder> for P4 3-groff is faling
11:44 < rxr> aha
11:45 < blindcoder> don't remember what regressions that coused, but I think it was xfree
11:45 < blindcoder> anyway, I'm off to lunch
11:45 < rxr> cu
11:58 < rxr> on osnews.com this article http://navindra.blogspot.com/2004/10/kde-dot-news-ext3s-miserable-failure.html made it into the news
11:58 < rxr> KDE Dot News
11:58 < rxr> Zope
11:58 < rxr> tar
11:58 < rxr> tar.gz
11:58 < rxr> ext3
11:58 < rxr> ReiserFS
11:58 < rxr> Directories
11:58 < rxr> -
11:58 < rxr> 30M
11:58 < rxr> 700K
11:58 < rxr> 200M
11:58 < rxr> 200K
11:58 < rxr> Full Monty
11:58 < rxr> 260M
11:58 < rxr> 270M
11:58 < rxr> 60M
11:58 < rxr> 930M
11:58 < rxr> 700M
11:59 < rxr> oh - shit - why can't cu'n paste tables work out better ...
11:59 < rxr> anyway look at the table on the arcticle's end ...
11:59 < rxr> 700k fot tat.gz vs. 200k on reiserfs - this guy should not believe that hisself ...
12:00 < rxr> as much as I favour reiser*fs, some huge directory structure will not be amaller than a tar.gz2 ....
12:02 < rxr> 954 builds total, 321 completed fine, 0 with errors.
12:09 < CIA-9> rene * r4399 /trunk/package/kde/kdelibs/kdelibs.conf: * substituted stray echo in the kdelibs.conf to echo_status
12:12 < CIA-9> rene * r4400 /trunk/package/x11/ (5 files in 3 dirs):
12:12 < CIA-9> * additions to r4396 and r4387 - postlinuc gatos* work - I must not
12:12 < CIA-9> forget to svn add files ;-)
12:20 < CIA-9> rene * r4401 /trunk/scripts/functions:
12:20 < CIA-9> * improved build_package to return quickly if a sub package set's
12:20 < CIA-9> custmain to "true" to not build anything anyways (this hides
12:20 < CIA-9> build errors due to missing downloads of such packages) - untested
12:30 < blindcoder> rehi
12:31 < rxr> damn - yet another one complaining I'm not reachable at rene@rocklinux.org ...
12:31 < rxr> clifford is such a nice guy
12:34 < rxr> oh my god: http://www.rocklinux.net/lurker/message/20041018.075400.6d5e4a75.html
12:35 < valentin> subject ?
12:40 < valentin> cu
12:46 < rxr> just some - plesae gimme a submaster patch for a one line change ...
13:08 < rxr> jsaw: netpbm does not like the new libtiff ... - undefined references :-(
13:09 < rxr> hm - and cinepaint errors out:
13:09 < rxr> checking for TIFF support... checking for TIFFOpen in -ltiff... no
13:09 < rxr> configure: error: *** TIFF dev lib not found ***
13:09 < rxr> this are the two only build errors I right now have ...
13:22 -!- nzg [~tschmidt@pD95F830D.dip.t-dialin.net] has joined #t2
13:23 < nzg> moin
13:31 < rxr> mon nzg
13:31 < nzg> moin rxr
14:04 < rxr> for people with time for bug hunting:
14:04 < rxr> [5] graphic/cinepaint [5] x11/netpbm
14:04 < rxr> [5] e17/efsd
14:04 < rxr> 955 builds total, 389 completed fine, 3 with errors.
14:10 < mnemoc_> moin
14:19 < rxr> hi mnemoc_
14:20 < mnemoc_> hi rxr
14:21 < rxr> if someone here is bored they could check links on:
14:21 < rxr> http://www.exactcode.de/
14:21 < rxr> and redirects on:
14:21 < rxr> http://www.rocklinux-consulting.de/
14:23 -!- nzg [~tschmidt@pD95F830D.dip.t-dialin.net] has quit [Read error: 110 (Connection timed out)]
14:24 < mnemoc_> what will you do with that domain?
14:26 < rxr> let it phase out over the next 11 payed months or so ...
14:26 < rxr> th said he might take it over then - just to have it around in case ...
14:26 < rxr> and not let it domain grab it ...
14:29 < mnemoc_> aha
15:24 < jsaw> ping rxr
15:25 -!- nzg [~tschmidt@p508EAAAB.dip.t-dialin.net] has joined #t2
15:35 < daja77> jsaw: were you pinging because of that create-errlist thing?
15:42 < jsaw> nope, a question in the source directory, why he uses strdup and not strdupa in Curl.hh
15:43 < daja77> pingin me,i meant
15:43 < jsaw> oh, wait, lemma think..
15:44 < jsaw> yes, I wanted to give u the answer here
15:44 < daja77> ah ok
15:45 < daja77> i could try that modification and send it to sm if you like
16:20 -!- Netsplit kornbluth.freenode.net <-> irc.freenode.net quits: mnemoc_
16:20 -!- Netsplit over, joins: mnemoc_
16:43 < jsaw> re
16:45 < jsaw> rxr: please add a coding style and some more information about what can be used and what not (e.g. why do you use strdup instead of strdupa in Curl.hh - no gnu extensions?). I'd like to contribute to the UpdateList a bit if I may.
16:46 < jsaw> *AAAAHHHH*
16:46 < jsaw> libtiff: usr/lib/libtiff
16:46 < jsaw> libtiff: usr/lib/libtiff.3
16:46 < jsaw> libtiff: usr/lib/libtiff.3.7.0
16:47 < jsaw> somebody wonders why cinepaint can't find the lib?
17:15 -!- rene [~rene@port-212-202-42-20.dynamic.qsc.de] has joined #t2
17:16 -!- rene [~rene@port-212-202-42-20.dynamic.qsc.de] has quit [Client Quit]
17:25 -!- nzg [~tschmidt@p508EAAAB.dip.t-dialin.net] has quit ["Verlassend"]
17:29 < CIA-9> jsaw * r4402 /trunk/package/graphic/libtiff/libtiff.conf:
17:29 < CIA-9> * recreate libtools in libtiff -> fixes installation
17:29 < CIA-9> of libraries (cinepaint just built)
17:35 < CIA-9> jsaw * r4403 /trunk/package/gnome2/glib/glib.desc: * update glib (2.4.6 -> 2.4.7)
17:35 < CIA-9> jsaw * r4404 /trunk/package/gnome2/gtk+/ (gtk+-2.4.9-xpm_ico_secure.patch gtk+.desc):
17:35 < CIA-9> * update gtk+ (2.4.9 -> 2.4.13)
17:35 < CIA-9> * remove patch that does not appy anymore
17:36 < CIA-9> jsaw * r4405 /trunk/package/x11/tcltk/:
17:36 < CIA-9> * remove disabled package tcltk, it's split into tcl and tk
17:36 < CIA-9> since a long time.
17:49 < CIA-9> jsaw * r4406 /trunk/package/network/iproute2/inc-order-fix.patch: * hotfix for iproute2 (taken over from rock)
17:50 < CIA-9> jsaw * r4407 /trunk/package/audio/alsaplayer/: * one alsaplayer is enough (is in multimedia/)
17:57 < CIA-9> jsaw * r4408 /trunk/package/x11/netpbm/ (netpbm.conf netpbm.desc shared_files-fix.patch):
17:57 < CIA-9> * update netpbm (10.12 -> 10.25)
17:57 < CIA-9> * updated shared_files-fix.patch: I have no idea what
17:57 < CIA-9> I've done there, but flist seems ok -> would be nice if
17:57 < CIA-9> somebody double checks
19:25 < Ragnar0k> hi
19:40 -!- BoS_ [~BoS@dialin-212-144-087-223.arcor-ip.net] has joined #t2
19:48 -!- BoS [~BoS@dialin-145-254-077-077.arcor-ip.net] has quit [Read error: 110 (Connection timed out)]
20:26 -!- Ragnar0k [loki@D8FA4DE6.ptr.dia.nextlink.net] has quit ["Reboot"]
21:41 < rxr> jsaw: valentin did the Curl stuff - no idea why he uses old C stuff ..
21:41 < rxr> real C++ STL stuff is preferred ...
21:43 < valentin> rxr: how ?
21:44 < valentin> can you tell me how to use mkstemp with std::string ?
21:44 < valentin> and why should i use strdupa when duplicating a constant ?
21:46 < valentin> btw: whats the difference between strdupa(ptr) and strdup(ptr) ?
21:46 < valentin> ah, i see, alloca
21:47 < valentin> i think this does not matter here
21:47 < valentin> jsaw: the updatelist just generates a list
21:47 < valentin> the evaluation should be done by an external program
21:48 < valentin> but you are free to invent smarter patterns for update recognition
22:26 < jsaw> re
22:27 < jsaw> valentin: you can make std::string create a copy - the only thing is, I do not remember how
22:28 < jsaw> clone() or so?
22:28 < valentin> a=b
22:28 < valentin> but this is not the problem
22:28 < jsaw> nope, I mean a writable copy.
22:28 < jsaw> what do you mean?
22:29 < valentin> like c_str() but returning a copied char * ?
22:29 < valentin> the problem is that it is not possible to overwrite the internal buffer of a std::string
22:30 < jsaw> yes. If you do std::string s="x", you get a non-writable string.
22:30 < jsaw> really?
22:30 < jsaw> I thought begin() gives the pointer to the memory location.
22:31 < jsaw> And also c_str(), but this extends the string so it gets zero terminated - overhead...
22:31 * valentin checking that
22:32 < jsaw> so probaly doing std::string s("x", 1) and the using c_str() will create the writable copy... (/me not doing so often C++ anymore, forgot it... :(
22:33 < valentin> but c_str is const
22:33 < jsaw> wait a second, I have to read the header.
22:34 < valentin> and begin just gives you an iterator
22:34 < valentin> you can use [] to access the items
22:35 < valentin> c_str is const according to the ISO std
22:36 < valentin> and i would not rely on implementation dependend internal representations
22:36 < jsaw> _M_data() returns the pointer to the data.
22:36 < valentin> public ?
22:36 < jsaw> So it is safe to cast the result of c_str() to (char *).
22:37 < valentin> this is not defined in the standard
22:37 < valentin> and i would not like to mess around with c++ hacks
22:38 < jsaw> _CharT*
22:38 < jsaw> _M_data() const
22:38 < jsaw> { return _M_dataplus._M_p; }
22:38 < jsaw> const _CharT*
22:38 < jsaw> c_str() const
22:38 < jsaw> {
22:38 < jsaw> [snip]
22:38 < jsaw> return _M_data();
22:38 < jsaw> }
22:38 < jsaw> what's the problem?
22:39 < valentin> because this relys on that specific stl implementation
22:40 < valentin> and why would a string hack("tmp/xXXXX"); mkstmp((char*) hack.c)str
22:40 < valentin> ...
22:41 < valentin> better than the plain c solution
22:42 < jsaw> allocating space, copy, (mkstemp), assign std::string, std::string must be forced to create copy because string is freed without std::string knowning.
22:42 < valentin> what is the problem ?
22:42 < valentin> there are four lines
22:42 < jsaw> double allocation
22:43 < jsaw> more calls than necessary, but hey stop:
22:43 < valentin> char* templ=strdup("/tmp/curl_downloadXXXXXX");
22:43 < valentin> mkstemp(templ);
22:43 < valentin> filename=templ;
22:43 < valentin> free(templ);
22:43 < jsaw> I just wanted to ask.
22:43 < jsaw> (these four lines create the allocation space for that string twice, and they copy it twice!)
22:44 < valentin> this probably slows the download down ...
22:44 < valentin> ,)
22:44 < valentin> just messing around in std::strings internal buffers is not very wise
22:45 < jsaw> in the end you are right. But I'm not that sure if filename=templ; always creates an internal copy (I mean you were talking about other STL's)
22:45 < valentin> yes it does
22:45 < valentin> and i am anoyed by it
22:46 < valentin> but i have not designed mkstemp
22:46 < valentin> this function sucks
22:46 < jsaw> with gcc, yes, but u can test stlport (just to make sure :)
22:46 < jsaw> brb (coffee)
22:46 < valentin> ok
22:48 < valentin> if the double copy bothers you we could use a char* filename instead of a std::string filename
22:49 < valentin> by the way: calling mkstemp in the constructor is not very clever anyway
22:49 < valentin> i should call it in Download ()
22:53 < valentin> the curl wrapper is preliminary implementation anyway
22:53 < valentin> i would like to use curl.h functions instead of calling system(curl ....)
22:59 < jsaw> re
23:01 < jsaw> valentin, don't forget, I'm just asking - mainly because I do not know the best solution myself. I usually tend to try to find the optimal solution while I'm not able to understand the simple solution :)
23:02 < jsaw> actually I should s/:)/:(/
23:03 < valentin> jsaw: and i'm not saying my solution is optimal
23:03 < jsaw> good.
23:03 < jsaw> so we agree,
23:03 < valentin> ok
23:04 < jsaw> I was just getting nervous if I'm upsetting you
23:04 < valentin> nope
23:04 < jsaw> does updatelist understand the cv-* thing?
23:04 < valentin> i think these four lines of code are not so important to discus about for more than 20 min :)
23:05 < jsaw> (/me scientist -> me loves to discuss just for fun, sorry)
23:06 < jsaw> (meine Frau rauft sich schon die Haare)
23:06 < valentin> then let us discuss how to procced with updatelist for fun
23:06 < jsaw> does it understand CV- tags?
23:07 < valentin> rxr decided to kick CV- tags
23:07 < jsaw> hm.
23:08 < valentin> but basicaly CV- tags would be no problem if we had them
23:09 < jsaw> I'm mainly thinking about path, which are of this form: some-path/${major}.${minor}/another-path/package-${major}.${minor}.${tiny}\(.${patchlevel\)?.tar.gz
23:10 < jsaw> would it make sense to detect this?
23:10 < jsaw> (would make life sometimes easier with gnome - if I don't have to start a browser each time I want to check the version.
23:10 < jsaw> )
23:11 -!- junix_ [~junix@pD9E75E2D.dip.t-dialin.net] has quit [Read error: 104 (Connection reset by peer)]
23:11 < valentin> jep
23:12 < jsaw> jep? to the sentence in brackets, or the one before?
23:12 < valentin> to both
23:13 < valentin> but this will be fun to implement ...
23:14 < jsaw> okay, so my proposal is this: 1. make UpdateList understand if only package name is given as argument, 2. slowly try to train it to guess the naming scheme given above.
23:16 < valentin> you are talking about training, so you want a neuronal net ? *G*
23:16 < valentin> but can you elaborate on 1) ?
23:17 < jsaw> hehe
23:17 < jsaw> yes. just have to understand the DirIterator.
23:19 < valentin> sth like UpdateList pkgname instead of UpdateList descfile ?
23:19 < jsaw> yes, but both.
23:27 < jsaw> 1301 builds total, 696 completed fine, 22 with errors
23:28 < jsaw> valentin: which editor do you use, and what coding style is it?
23:32 < valentin> i use xemacs
23:32 < jsaw> style?
23:33 < valentin> coding style like it is
23:33 < valentin> do not know the name
23:33 < jsaw> then it's gnu iirc
23:33 < valentin> normal xemacs c++-mode
23:33 < jsaw> I usually use stroustrup
23:34 < valentin> just ident as you like, identation can easily be changed later
23:38 < jsaw> rxr: lim(t->2.1.0) err(t) = $pkgs = 0 ????
23:52 < valentin> rxr is away sleeping ..
23:53 < valentin> i think the = $pkgs is too much
23:53 < valentin> or he means err / $pkgs
23:53 < valentin> or whatever
--- Log closed Mon Oct 18 00:00:24 2004