Camelia, the Perl 6 bug

IRC log for #cdk, 2009-04-27

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

All times shown according to UTC.

Time Nick Message
05:05 bag joined #cdk
05:35 jbrefort joined #cdk
05:36 egonw joined #cdk
05:47 rguha joined #cdk
05:56 egonw hi rguha
05:56 rguha egonw: hi
05:57 egonw can you email the patches?
05:57 egonw I don't seem able to download the first from SF
05:57 egonw with some connection broken error
05:57 rguha one sec
05:58 egonw sure
05:58 egonw (you're up late, not)
05:58 rguha sent
05:58 rguha yes :)
05:58 egonw thanx
05:59 egonw only one file?
05:59 rguha yes
06:00 rguha I can't make the original pcore patch since I don't have that branch anymore
06:00 egonw no problem
06:00 rguha the idea was to merge in the pcore branch and then apply this latest patch
06:00 egonw btw, agreed on that import issue
06:01 rguha aah, ok
06:01 rguha merging in the threadtag branch (trying to)
06:01 egonw 1.2.x?
06:02 egonw that's where it needs to go
06:02 rguha yes
06:02 egonw I think we only have one review yet...
06:02 egonw not?
06:02 egonw it is new code
06:02 egonw and not s bug fix
06:02 egonw I'll ask stefan to look at it
06:02 rguha which one? pubchem fp's?
06:02 egonw no threadtag
06:02 rguha oh right - but it's small enough that 1 review seems fine
06:03 rguha and it desn't affect the main API. It's a taglet
06:03 rguha i saw Thorsten recommend it as well (though not a strict code review)
06:04 egonw well, the did a good job at #cdkws2009
06:04 egonw reviewed 31 classes
06:04 egonw see wiki
06:04 rguha yes, noticed that
06:06 rguha so do we wait for a code review of the threadtag patch? or just commit?
06:08 egonw wait
06:08 egonw since it will trigger commits in the code sources itself
06:10 rguha ok
06:10 rguha btw, did the cdk.githash patch go through?
06:11 egonw pcore patch applied
06:11 egonw don't think I applied that yet
06:11 rguha thanks
06:11 egonw did I review it already? (or someone else?)
06:11 rguha yes, miguel reviewed it
06:12 rguha I sent a second patch in response to that to cdk-devel (to be applied after the original patch)
06:20 egonw rguha: I cannot apply them
06:20 egonw the fail on master and cdk-1.2.x
06:21 rguha can you post the error?
06:22 egonw on master I get:
06:22 egonw error: src/main/org/openscience/cdk/qsar/desc​riptors/molecular/ChiIndexUtils.java: patch does not apply
06:22 egonw on trunk the EStateAtomTypeMatcher stuff
06:22 rguha this is the cdk.githash patch?
06:22 egonw yes
06:22 egonw git am ~/Desktop/0001-Converted-cdk.svnr​ev-tags-to-cdk.githash-tags.patch
06:22 rguha aargh. let me regenerate it
06:23 egonw please on cdk-1.2.x
06:25 sneumann joined #cdk
06:32 rguha mailed the 2 patches directly to you
06:35 egonw ack
06:37 edrin joined #cdk
06:38 egonw hi edrin
06:40 rguha off to sleep. if there any problems with the githash patch, let me know
06:41 rguha btw, can you delete cdk.1.2.x of SF - I pushed it there by mistake
06:42 egonw ah, it was you :)
06:42 egonw I already was wondering how I got that there :)
06:42 rguha :)
06:42 egonw see the last wiki question
06:42 egonw please try yourself
06:43 egonw rguha: please review this (small) patch: https://sourceforge.net/supp​ort/tracker.php?aid=1864371
06:44 egonw I am setting up a git branch for it
06:46 rguha ok, deleted
06:48 rguha is the new patch for 1.2.x or master?
06:49 egonw 1.2.x
06:49 egonw if you mean the patch I was last talking about
06:49 rguha yes, the CDKException patch
06:49 egonw yes, 1.2.x
06:49 egonw about to send a branch to my pele repos
06:50 rguha let me know when to pull
06:53 rojasm joined #cdk
06:54 rojasm moin
06:54 zarah ni hao rojasm
06:55 egonw rojasm: do you got time to review to small patches
06:57 rojasm yeah
07:01 Gpox joined #cdk
07:01 rguha goodnight
07:04 egonw rojasm: patches forwarded
07:05 rojasm I am reading now. http://sourceforge.net/tracker/​?group_id=20024&atid=320024
07:05 zarah rojasm's link is also http://tinyurl.com/crwodv
07:05 rojasm but more of them I don't know the algorithm behind
07:10 egonw what algorithm?
07:17 rojasm Maybe I am not enough concentrated.
07:17 rojasm The idea with patches is:
07:17 rojasm 1. using tracker patch webpage functionality for open a patch and find the best solution between all developer
07:17 rojasm 2. create a branch to modify the problem.
07:17 rojasm 3. merge the branch to the trunk
07:19 egonw yes
07:21 egonw the two patches I sent are a follow up on comments you gave earlier
07:21 egonw sent to you Leiden Univ account
07:29 rojasm received
07:39 rojasm Some thing that I don't understand is: The source for svn and git will be composed from different codes? I mean why is eliminated the class CDKSVNTaglet. When in SVN will be still used?
07:42 egonw SVN is no longer used for the CDK library
07:45 rojasm ok
07:52 rojasm It looks fine for me. how I should report that?
07:52 egonw ideally, you add signoffs
07:53 egonw do you have a local git branch with the two patches?
07:54 rojasm I am importing now the git cdk project with eclipse
07:54 rojasm it needs some minutes still
07:54 egonw yeah, fine :)
07:54 egonw no rush
07:55 * rojasm using eclipse git-plugin to download cdk-git project
07:57 egonw rojasm++
07:58 rojasm oo my eclipse gets out of memory....
08:01 jonalv joined #cdk
08:04 egonw egit--
08:04 jonalv what's the difference between mdl and mol files?
08:05 egonw mdl is a company
08:07 jonalv egonw: uhm, okey. But are the file formats related?
08:07 * jonalv has no idea...
08:07 egonw mdl is not a file format
08:07 egonw "mdl molfile" is a format
08:08 jonalv eh okey, so what it the difference btween "mdl molfile" and mol files then?
08:08 shk3 joined #cdk
08:19 egonw jonalv: nothing, if you refer to the same thing
08:19 egonw what do you mean with 'mol file' ?
08:19 jonalv egonw: uhm what did you mean with mol file in tha bug repport?
08:20 egonw prob MDL molfile :)
08:20 jonalv egonw: you used the two things as different things. I relaised that I don't know about both so I want to learn...
08:24 egonw I think that was unintentional
08:26 maclean joined #cdk
08:39 egonw bbl
08:42 rojasm egonw at the moment there are only the master, isn't it?
08:42 rojasm not branches
08:45 maclean I think that branches are local
08:47 rojasm ok now I see. With eclipse-git I can see Local Branches and Remote branches
08:47 rojasm in remote branches we have cdk-1.2.x and the master
08:48 maclean eclipse-git?
08:48 rojasm yes
08:48 maclean I mean : which eclipse plugin do you use?
08:48 maclean I tried one, but it wasn't very helpful.
08:49 rojasm I don't know if it is good!
08:49 rojasm I am install the spearce.org Java Git/ Eclipse Git, feature id org.spearce.egit
08:50 rojasm I am using I mean
08:52 rojasm it has the History view which it shows graphically all the commits and merges
09:26 maclean joined #cdk
09:27 maclean ah right, cool thanks. I'll give it a go.
09:30 rojasm After I update the branch cdk.1.2.x and compile with  the ant javacc
09:31 rojasm Git complains that I have new class NomParser.java, blabla
09:32 rojasm how show I go ahead. In subversion was not a problem
09:45 maclean ah, sorry rojasm, I'm not sure.
09:46 maclean Is it that NomParser is a class generated by the javacc process?
09:47 rojasm I know that cdk create some java class using the ant javacc which are specific of system
09:47 rojasm so needed step before we can work with cdk library
09:48 rojasm now I read that with git we can set some file which should be ignore calling the command .gitignore
09:48 rojasm egonw: should it this file include in cdk
09:50 rojasm about the new files created
09:56 maclean I would have thought that the generated java files were already in .gitignore.
09:57 rojasm not they are not
09:57 rojasm I sent just now a email to dev-list about that
09:57 maclean I saw :)
10:16 egonw joined #cdk
10:21 rojasm egonw I report some problems with created files using ant javacc
10:22 rojasm other question the branches which are working the people is the master or the cdk-1.2.x
10:22 rojasm I mean from what of both I should create local branch
10:23 rojasm to add the patches
10:25 rojasm egonw sorry more questions. Importing the patch are you using the function:
10:25 rojasm $ git am -3 patches.mbox
10:28 CIA-58 cdk: shk3 * r14571 /jchempaint/trunk/build.xml: added all external jars (inchi still has library problems)
10:36 shk3 joined #cdk
11:10 egonw joined #cdk
11:41 maclean joined #cdk
11:42 egonw joined #cdk
11:58 egonw Gpox: ping
12:01 Gpox egonw: pong
12:01 egonw I want to upgrade CDK to 1.2.1
12:02 egonw but
12:02 egonw I notice many patches in cdk-externals
12:04 egonw shk3: ping
12:05 egonw Gpox, shk3: can you please go through the list of patches applied to cdk-externals/trunk and check if yours have been applied to jchempaint-primary or cdk-1.2.x ?
12:05 egonw I cannot upgrade to CDK 1.2.1 now
12:07 shk3 sorry
12:07 shk3 what is cdk-externals?
12:07 egonw git-svn/bioclipse2/cdk-externals/trunk
12:08 egonw shk3, Gpox: can we agree on the requirement to add to the commit message how the patch is submitted upstream?
12:08 egonw shk3, Gpox: so that anyone can easily check if all patches are applied upstream?
12:08 egonw upstream := jchempaint-primary | cdk-1.2.x
12:09 shk3 I have no idea what you are talking about
12:09 maclean so, commit messages should have one of these two tags in them?
12:10 egonw no, patches that go into cdk-externals *must* get applied upstream
12:10 egonw and right now it is impossible to see which ones have been reported upstream
12:11 egonw I don't like to see Bioclipse have a CDK fork
12:11 egonw the CDK in Bioclipse SVN should only be a copy of upstream
12:12 shk3 are you talking about if stuff commited to bioclipse svn is in cdk repository?
12:13 egonw I am talking about the patches made to bioclipse SVN in cdk-externals
12:13 egonw ...
12:13 egonw of which I want to know if that got applied upstream
12:13 egonw upstream := CDK git / CDK svn
12:13 egonw depending on cdk-1-2-x or jchempaint-primary respectively
12:14 shk3 ah ok
12:14 egonw shk3, Gpox and me made patches
12:14 shk3 i did not realize cdk externals is bioclipse
12:14 egonw I know which of mine have been applied upstream
12:14 egonw but I do not know about those of shk3 and Gpox
12:14 shk3 from me, only the modelbuilder stuff is pending
12:14 shk3 I am going to make a new patch for that
12:15 egonw Gpox: can you please check too?
12:15 Gpox will do
12:17 egonw thanx
12:17 egonw what about putting info on upstream reports in the commit message?
12:17 egonw can we agree on that?
12:17 egonw then I don't have to bother you next time CDK needs upgrading
12:20 egonw shk3, Gpox: ?
12:21 shk3 put which messages where?
12:21 maclean but what _exactly_ should be put in the message...
12:21 shk3 sorry, I don't understand?
12:21 egonw when you apply a patch to cdk-externals, you need to report that patch upstream
12:21 shk3 that's what I always did
12:22 egonw add to the commit message of the cdk-externals patch, *where* you reported the patch upstream
12:22 egonw shk3: not that I can see
12:22 egonw bug #860 fixed again (seems by some changes it krept up again)
12:22 egonw
12:22 egonw no mention there *where* you reported the patch upstream
12:22 egonw the things is...
12:22 shk3 I meant I did it
12:22 shk3 but did not report
12:22 shk3 yes
12:22 egonw right
12:22 egonw so, if you mention where in the commit message above
12:23 egonw then
12:23 egonw I can check if the patch has been applied upstream
12:23 egonw without having to annoy you with these stupid maintainance issues
12:23 egonw it would allow me to figure out myself what is and has not been applied yet
12:24 egonw so, what I want to propose is to just add a line in that commit message pointing to the CDK patch trakcer
12:24 egonw or even SVN commit number if already applied
12:25 shk3 will do so
12:25 egonw Gpox: you are happy with such an agreement too?
12:25 Gpox yes
12:26 egonw great, thanx guys!
12:26 egonw for now...
12:26 egonw shall I upgrade to 1.2.1 and reapply the patches one by one?
12:26 egonw so, excluding .control and .render
12:27 egonw .1 has a few relevant bug fixes
12:27 egonw e.g. in SMARTS
12:27 shk3 I think that's the only possibility if you need an update
12:37 egonw shk3: ping... do you want to learn some more git magic?
12:37 egonw in particular how to collapse patches?
12:38 egonw or, better worded: how to merge patches into a single patch?
12:38 shk3 reluctant
12:38 egonw :)
12:38 egonw well, just let me know
12:38 shk3 gilliean tells me to learn it
12:38 shk3 so go ahead
12:38 maclean there's a lot to learn with git...
12:39 egonw ok, in your branch, with those four patches
12:39 egonw try this in the command line:
12:39 egonw git rebase -i HEAD~5
12:39 egonw that will open a plain text editor listing the last five patches
12:40 egonw four of which are yours...
12:40 egonw be careful
12:40 egonw as this is powerfull stuff
12:40 egonw removing a complete line, removed the full patch
12:40 egonw the pick indicates: preserve this patch
12:40 egonw edit allows you to make corrections to that patch
12:41 egonw but what I wanted to tell you about is the 's' option
12:41 egonw the first line is the oldest patch
12:41 egonw if the last four are yours
12:41 egonw you can mark the last *tree* as squash
12:41 egonw which will merge them with the last before that...
12:41 egonw the oldest of your four patches
12:41 egonw this will create a single new patch
12:42 egonw merging your four patches
12:42 egonw this is not something you can do when people base their work on your branch
12:42 egonw but suitable for applying reviewer commetns...
12:43 egonw as the final apply of the patch will typically use a fresh copy of the full set of patches
12:43 egonw the reason why I tell this, and Mark might find it interesting too...
12:44 egonw is that you merge replies to reviewers feedback at the location where the change is made...
12:44 egonw in your case:
12:45 egonw fix the 'Description of parameter' where you changed Exception into CDKException
12:45 egonw no need to do this for your four patches, but just as something to keep in mind in the future
12:46 maclean egonw : one thing that occurred to me...
12:47 egonw yes...
12:47 maclean with this distributed system
12:47 maclean what happens if changes get applied out of order
12:47 maclean so If I pull from developer A, and then developer C
12:48 egonw that's why I mentioned something about others building there work on your branch...
12:48 maclean but A pulls from C and D, and merges happen at various points...
12:48 egonw so, if you published something, you typically should not do this...
12:48 egonw because that will mess up things seriously...
12:48 maclean well indeed.
12:49 egonw but since shk3's patch only goes here, and I will apply a fresh copy anyway, it is not really a problem
12:49 maclean but this is why relying on knowing where people's repositories are is dangerous
12:49 egonw using this approach, one can make cleaner patches
12:49 egonw and I anticipate we might use this when makeing jcp-prim a patcjh to cdk-1.2.x later
12:50 maclean okay.
12:50 maclean oh, about the jcp-primary->1.2.x merge
12:50 maclean could it be done in two stages?
12:51 maclean with the renderer package first, then the controller later?
12:51 maclean as the former is planned to be more stable than the latter
12:51 egonw maclean: please see the CDK wiki page about it
12:51 rojasm question. why is still in git branch the cdk-1.2.x project when this was already merged 3 days ago (2009-04-24). Is there some reason?
12:51 maclean and more people want to use the ...oh, ok
12:51 egonw we are still planning how to apply what in which order
12:51 egonw maclean: please add your ideas to that wiki page
12:51 egonw but, yes...
12:51 maclean ok
12:52 egonw renderer first, then controller
12:52 egonw even better:
12:52 egonw for each: interfaces first
12:52 egonw then implementation
12:52 egonw and so fort
12:52 egonw even before that: patches outside .render and .control
12:52 maclean yes, that makes sense
12:52 egonw rojasm: because cdk-1.2.x is still open for bug fixes
12:53 rojasm achso
12:53 egonw rojasm: and those merge commits indicate when those bug fixes were ported to master
12:54 rojasm ok! now I see
12:55 shk3 egonw: how to do a replace in git?
12:56 jpansanel moin
12:56 zarah oh hai jpansanel
12:56 egonw hi jpansanel
12:56 egonw shk3: what kind of replace?
12:56 shk3 replace a changed file with the repository version
12:56 egonw git reset HEAD <FILE>
12:56 egonw or something like that
13:04 rojasm how is the way dong a patch? each commit should be in a different patch or we can put several commits in the same patch.
13:04 rojasm with the command: git diff master..experiment > experiment.patch
13:05 egonw better:
13:05 egonw git format-patch master..experiment
13:05 egonw rojasm: still better:
13:06 egonw just publish your branch online
13:29 rojasm to publish I am doing that:
13:29 rojasm git branch miguelrojasch-formulaupdate
13:29 rojasm git checkout miguelrojasch-formulaupdate
13:29 rojasm ## some modifications
13:29 rojasm git commit -a -m "this modifications"
13:29 rojasm git push public miguelrojasch-formula
13:29 rojasm ++++++
13:29 rojasm is It correct?
13:33 egonw yes, where public is something...
13:33 egonw could be a your bare repos on leidenuniv.nl/~rojasm/cdk.git/
13:34 egonw or GitHub
13:34 rojasm mmm. or is it not possible in Sourceforge?
13:35 egonw no
13:35 rojasm ok
13:37 rojasm why not?
13:37 egonw because there is no direct need
13:38 egonw and I like to see the official branches on there only
13:38 egonw but if you cannot find another place, let me know, then we work something out...
13:38 egonw you can always fall back to emailing formatted patches
13:39 rojasm ok I will put in github then
13:55 CIA-58 cdk: egonw * r14572 /cdk-xws/trunk/ (xws-cdk-services/.classpath xws-qsar-services/.classpath): Upgraded to CDK 1.2.1
13:55 CIA-58 cdk: egonw * r14573 /cdk-xws/trunk/xws-qsar-se​rvices/jar/cdk-1.1.1.jar: Removed jar form SVN
13:56 CIA-58 cdk: egonw * r14574 /cdk-xws/trunk/xws-qsar-services/src/org/​openscience/cdk/xws/services/descriptor/ (BCUT.java TPSA.java XLogP.java): Marked functions as ASYNC
15:21 rajarshi joined #cdk
15:21 rajarshi egonw, do you think that when e commit a series of related patches from a contributor, it'd be useful to package themn into one patch?
15:27 egonw depends
15:28 egonw keeping patches somewhat split up makes it easier to review
15:28 egonw and when you squash
15:28 egonw you loose Author: info
15:28 egonw (as I discovered when squashing shk3's patches earlier today :)
15:28 rajarshi aah, ok
15:28 rajarshi i was just wondering about pauls patches - 14 commits, which are all very closely related
15:28 rajarshi anyway, pushed to 1.2.x and master on sf
15:29 egonw ok, thanx
15:30 rajarshi did the githash tag patch makeit through yet?
15:30 egonw rojasm: ping
15:32 egonw rajarshi: please ask rojasm
15:32 egonw when he is happy, I am happy
15:33 rojasm egonw pong
15:33 rajarshi ok :)
15:34 egonw rojasm: did you look at the githash javadoc tag patch?
15:34 rajarshi rojasm, did you get to see the latest patch for the cdk.githash tag that addressed your comment wrt renaming the taglet class?
15:34 egonw rajarshi: yes, forwarded him your email
15:34 rojasm not sorry!
15:34 rojasm yes I sow
15:36 rajarshi does it look OK?
15:36 rojasm yes! I looked the file with a normal editor
15:37 rojasm sorry I didn't answer I wanted to used the functionality of the git patches to check the differences
15:37 rojasm But I didn't get to do it
15:37 rajarshi aah, ok
15:38 rojasm For me looks great know
15:39 rojasm only what I told egonw. That I thought subversion and git would run parallel and you should not remove the class for SVN
15:39 rojasm but egonw explained me that in principle SVN-cdk is opsolet!
15:40 rajarshi right - after tasting git, no reason to use svn :)
15:40 egonw at least for the CDK library, yes
15:41 egonw rojasm: do you want to signoff the patches?
15:41 rojasm yes!
15:42 egonw rojasm: OK, then I'll await your GutHub branch :)
15:42 egonw rojasm: you know how to signoff commits?
15:42 rojasm not
15:42 rojasm but now I have to leave already :(
15:42 egonw ok, check this wiki page:
15:42 rojasm can we do tomorrow
15:43 egonw https://apps.sourceforge.net/mediawiki/c​dk/index.php?title=Development_with_Git
15:43 rojasm ok
15:43 egonw yes, fine
15:43 rojasm I wrote the page already
16:01 maclean joined #cdk
16:21 timvdm joined #cdk
17:11 shk3 joined #cdk
17:33 maclean @tell egonw : could the IIteratingChemObjectReader interface extend Iterable<ChemObject> ?
17:33 zarah Consider it noted.
17:33 maclean ta luv
17:35 maclean oh, wait, it extends Iterator.
17:35 maclean hmmm.
17:38 maclean @tell egonw never mind, I found out why not
17:38 zarah Consider it noted.
17:56 egonw joined #cdk
18:01 bag joined #cdk
18:10 edrin joined #cdk
19:14 edrin joined #cdk
19:48 * egonw heads to other room for some TV and then bed
19:48 egonw more CDK patching tomorrow
19:48 egonw I think we are catching up :)
23:23 timvdm joined #cdk

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