Camelia, the Perl 6 bug

IRC log for #cdk, 2008-02-17

| Channels | #cdk index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:08 alexandros_ joined #cdk
01:16 alexandros__ joined #cdk
01:23 alexandros joined #cdk
06:52 jbrefort joined #cdk
07:28 egonw joined #cdk
07:28 egonw moin
07:31 CIA-31 cdk: egonw * r10148 /branches/cdk-1.0.x/src/org/openscience/cdk/ (6 files in 4 dirs): Added missing Jumbo52 requirements, and a conditionally loading of the Jumbo52 test suite
07:36 man-di egonw: good morning
07:36 egonw hi man-di
07:36 CIA-31 cdk: egonw * r10149 /branches/cdk-1.0.x/build.xml: Cleaned up a bit
07:39 egonw man-di: I realized a much better solution for the .jar lib names... instead of renaming them to match Debian's names, I will now use a Ant property...
07:39 egonw so, the ant.properties will contain lines like:
07:39 man-di I wanted to propose that from the start, sorry I forgot
07:39 egonw weka.jar=weka-3.5.7.jar
07:39 man-di other projects do this this way too and it works great
07:39 egonw no worries...
07:40 egonw well... stupid thing is...
07:40 egonw I did discuss this option in the past...
07:40 egonw with daniel? not sure...
07:40 man-di and you dismissed it?
07:41 egonw no, the other solution was less work...
07:41 egonw was too tired yesterday, I guess...
07:41 egonw but it has huge advantages...
07:41 egonw no renaming upon version updates, etc, etc, etc...
07:41 egonw just resetting some props...
07:42 man-di well, you should code defaults into build.xml
07:42 egonw yes:
07:42 egonw <property name="weka.jar" value="weja.jar"/>
07:42 man-di right
07:42 egonw actually, this should work for Debian too..
07:42 egonw as I just saw
07:42 egonw because there seems to be a sym link
07:42 egonw not all java packages do that...
07:42 man-di yes, we always have a symlink without version number
07:43 egonw jgrapht for example not
07:43 man-di if not, its a bug
07:43 egonw right...
07:43 egonw jgrapht only has an unversion one...
07:43 egonw well, my .deb of it...
07:43 egonw not sure if sid has a newer...
07:43 egonw mom
07:43 man-di yes, jgrapht is on my list to fix this
07:43 egonw no, don't think so...
07:44 man-di btw: does cdk work with newer jgrapht?
07:44 man-di it includes 0.6.0
07:44 egonw that's what we have too...
07:44 egonw in trunk
07:44 egonw let me check 1.0.z
07:44 man-di will it work with 0.7.3?
07:45 man-di I would like to updated jgrapht
07:45 egonw have not tried that yet...
07:45 man-di might be a TODO for next release for you. not for this release
07:45 egonw yeah
07:45 egonw let's get 1.0.2 into debian first
07:45 man-di yes
07:45 egonw will make a note about it...
08:07 egonw filed a bug report
08:07 egonw also noted that 1.0.x does not compile against the weka.jar in debian
08:07 egonw 3.5.7
08:07 egonw not sure what version 1.0.x was using...
08:09 egonw s/was/is/
08:10 man-di how hard is it to fix this?
08:11 egonw 12 compile errors...
08:11 egonw seems a class missing
08:11 egonw but it's just a minor part of the CDK...
08:11 egonw filed a bug report
08:11 egonw and will talk to Miguel who wrote this code about it
08:12 man-di 3.5.7 seems to be the latest version of weka
08:12 egonw not sure what version is in 1.0.x
08:13 egonw # CDK 1.0 does not compile against Weka 2.5.7, see SF bug report #1895295
08:13 egonw weka.jar=skip
08:13 man-di jumbo is not in Debian, I will work on it when cdk is in
08:13 CIA-31 cdk: egonw * r10150 /branches/cdk-1.0.x/ (build.xml src/org/openscience/cdk/libio/weka/Weka.java): Compiling weka depending code is now optional
08:13 man-di egonw: 3.5.7 not 2.5.7
08:13 egonw right
08:13 egonw ok, batik is up next
08:15 egonw ok, going to skip that too...
08:15 egonw cdk 1.0.x is using 1.5.1
08:15 egonw debian has 1.6
08:15 egonw but this is even less important
08:15 egonw SVG writing is only used by JChemPaint
08:15 egonw which is not even yet compiled
08:15 man-di debian is going to update to batik 1.7
08:16 egonw ok, better even wait until that is done...
08:16 man-di just to inform you
08:16 egonw this will involve not just compiling, but testing too...
08:16 egonw I will add a ant property compile.batik...
08:17 egonw testing for the right jars is trickier as 1.5 and 1.6 have different set ups...
08:18 egonw and I'll have ant.properties compile.batik=false
08:21 man-di okay
08:46 rojasm joined #cdk
08:52 egonw hi rojasm
09:06 CIA-31 cdk: egonw * r10151 /branches/cdk-1.0.x/ (3 files in 3 dirs): Now conditionally compiles Batik depending code
09:07 rojasm hi egonw
09:15 egonw rojasm: what weka version are you using in the CDK?
09:26 rojasm uff!
09:26 egonw anyway...
09:26 rojasm really I don't remenber
09:26 CIA-31 cdk: egonw * r10152 /branches/cdk-1.0.x/src/org/openscience/cdk/ (18 files in 4 dirs): Fixed weka-depending compiling of test-qsar
09:26 egonw rojasm: it does not compile against weka 3.5.7 in Debian unstable
09:26 egonw just that you are aware...
09:26 egonw filed a bug report about it
09:26 rojasm ok
09:27 rojasm I used always a java -jar
09:27 rojasm I wiil look If I find it
09:31 egonw no rush
09:44 CIA-31 cdk: egonw * r10153 /branches/cdk-1.0.x/build.xml: Cleaned up code, and added compile-info task to complement the already existing info task. The compile-info depends on the source code too, so runs runDoclet
09:46 CIA-31 cdk: egonw * r10154 /branches/cdk-1.0.x/build.xml: No need anymore to put .javafiles in the puredist
10:18 CIA-31 cdk: egonw * r10155 /branches/cdk-1.0.x/build.xml: Cleaned up deps a bit
10:26 CIA-31 cdk: egonw * r10156 /branches/cdk-1.0.x/ (5 files in 3 dirs): Optionally compile jniinchi code
10:33 egonw man-di: ok, running the pdebuild again
10:33 man-di great
10:36 egonw it's downloading 47.6MB of new .deb's
10:42 egonw man-di: I note one issue...
10:42 egonw the weka package seems to depend on sun-java6-jre
10:42 egonw another reason why the CDK should not compile the weka blob
10:44 egonw OK, changed my .pbuilderrc to only include main
10:44 egonw man-di: suggesting something from non-main is OK, right?
10:45 * egonw hopes his sid pbuild env understands that it should now only look in main
10:45 egonw and not contrib
10:48 egonw ok, it seems to not pull sun-java anymore
10:49 egonw ok, made progress
10:49 egonw but it fails...
10:49 egonw it gets to the compiling
10:49 egonw but does not seem to pick up the log4j jar...
10:54 egonw man-di: ping
11:00 egonw arghh... sometimes I just hate Ant
11:04 CIA-31 cdk: egonw * r10157 /branches/cdk-1.0.x/build.xml: Give feedback on the nowarn property (which does not seem to have any effect)
11:04 CIA-31 cdk: egonw * r10158 /branches/cdk-1.0.x/src/org/openscience/​cdk/test/inchi/InChIGeneratorTest.java: Added missing requirement
11:20 CIA-31 cdk: egonw * r10159 /branches/cdk-1.0.x/build.xml: Add the log4j.jar value as parameter
11:21 egonw ok, I think I solved the log4j problem
11:21 egonw debian has -15, while CDK had -13
11:21 egonw so pointing to the wrong jar
11:21 egonw ubuntu still has -13
11:26 egonw ah... of course...
11:26 egonw still still requires a patch of the .libdepends...
11:26 egonw man-di: around to teach me how to do patches properly in debian/ ?
11:27 * egonw starts googling...
11:30 man-di egonw: cdk uses simple patch files in debian/patches
11:30 man-di just put a patch file in there and be done
11:30 egonw the debian/rules on alioth will automatically pick it up?
11:30 man-di yes
11:31 egonw ok, excellent...
11:31 egonw one question...
11:31 man-di as weka depends on sun jdk we need to put cdk into contrib
11:31 egonw say we have cdk-1.0.1.9/debian
11:31 man-di I will take care of this
11:31 egonw I would create the patch from the cdk-1.0.1.9/ dir?
11:31 egonw no, cdk does not depend on it...
11:31 egonw part of the code could use it...
11:32 egonw but it is certainly not a compile dependency
11:32 egonw the fact that the orig.tgz has code that can only be compiled with stuff from contrib, makes the whole package contrib?
11:32 egonw even if one would not compile those bits?
11:32 man-di yes
11:32 egonw crap
11:33 man-di no. only if its compiled against, actually used
11:33 egonw ok
11:33 egonw so, not if we don't compile against it?
11:33 egonw because I am currently not doing that...
11:33 egonw that actually fails...
11:33 man-di so if we disbale weka this doest render it contrib
11:33 egonw because of weka version incompatability
11:33 man-di ah
11:33 egonw it's currently disabled...
11:34 egonw so it could go into main?
11:34 egonw my pbuilder is main only
11:34 man-di egonw: which is good ;-) but easy to change
11:34 man-di egonw: I have main-only and main/contrib/non-free pbuilders
11:35 egonw I'm aiming at main
11:36 man-di okay
11:36 man-di good to know
11:36 man-di batik ist contrib too
11:36 egonw right
11:36 egonw but is disabled too :)
11:37 man-di there are many Java packages in contrib
11:37 egonw yeah, I know
11:37 egonw I've monitored the move-to-main progress for a while
11:37 egonw I wish I had time to help out... but don't :(
11:37 man-di no problem, we will sort that out sooner or later
11:38 egonw anyway... that's why I aim at main... not to make it even worse :)
11:38 egonw when J1.7 is releaed... the problem will significantly reduce
11:38 egonw with speedy Swing
11:39 egonw which is blocking Jmol effectively
11:39 man-di we will see
11:39 egonw :)
11:39 man-di I dont see that yet
11:40 egonw I haven't followed J1.7 development much recently...
11:40 egonw see bits of icedtea
11:41 egonw ok, placed the patch in debian/patches
11:41 egonw and now rerunning pdebuild
11:41 egonw will have lunch now, and bbl
11:42 egonw with the results
11:42 * man-di moving outside to enjoy the weather, cu later
11:49 egonw it's working :)
11:49 egonw it now ran into a missing apache-cli jar issue :)
11:49 egonw that's in the applicatino module..
11:49 egonw one of the very last it tries to compile :)
12:08 egonw now I'm only hoping that 1.1 does not have API changes compared to 1.0
12:13 egonw yes, it compile!
12:15 egonw +s
12:15 egonw but fails, because of debian/ trouble ;)
12:15 egonw but nothing man-di can't solve
12:16 * egonw will now make a 1.0.1.9 release candidate for 1.0.2
12:16 egonw with puredist only :)
12:16 egonw so, not really a RC
13:59 dleidert joined #cdk
14:08 edrin joined #cdk

| Channels | #cdk index | Today | | Search | Google Search | Plain-Text | summary