From rubyforge at foemmel.com Sun Apr 1 16:18:28 2007 From: rubyforge at foemmel.com (Matthew Foemmel) Date: Sun, 1 Apr 2007 15:18:28 -0500 Subject: [Raven-devel] Minor 2.0 patches In-Reply-To: References: <460E99D3.3060401@herzumsoftware.com> Message-ID: Hi Matthieu, I think we should switch to svn:external. When we first set up that vendor brach, JRuby was a lot buggier than it is now, and I was thinking we'd be trying out a lot more changes locally before submitting them to the JRuby team. But it's pretty stable now, and as you say they've been quite responsive in terms of getting patches applied, so I think just using svn:external makes sense. Cheers, Foemmel On Mar 31, 2007, at 2:54 PM, Matthieu Riou wrote: > Hi Jesse, > > I've just applied your fixes, they all make sense. Thanks a lot! > > JRuby is imported as a vendor branch (see the vendor directory > under to repository root). The goal was to be able to patch the > sources directly if we needed to fix something, as JRuby is still > not quite there. However they tend to be pretty responsive and > apply patches almost immediately so maybe we could transition to > svn:external. What do you think Matthew? > > Cheers, > Matthieu > > On 3/31/07, jbouwman wrote: > Hello > > I've seen a few small issues on trunk, and had to make the following > changes to get it working for me: > > 1. In javac_task.rb, had to push @destdir onto the class_dirs > channel in > order for jar task to see it. Is this code changing? The comment > above > mentions this channel, so it seemed odd that it was missing. > > 2. In gem_task.rb, changed the :project attribute writer to :group, > which seems to be the code's 'mavenistic' intention. > > 3. Also in gem_task, changed the task name from 'gem', since it > collides > with a similarly named function in JRuby 0.9.8. Not sure what the best > name would be here. > > 4. in java_tasks.rb, change to ignore both ".xyz" and "xyz~" filenames > in raven/task directories on the load path, ir order to ignore emacs > backup files. > > I do think it makes sense to use an svn external to pull in jruby-- > I was > surprised not to see that, on looking at the sources for the first > time. I've attached diff output for the above things. > > > Thanks! > > > Jesse > > > > Index: raven/tasks/javac_task.rb > =================================================================== > --- raven/tasks/javac_task.rb (revision 182) > +++ raven/tasks/javac_task.rb (working copy) > @@ -87,6 +87,7 @@ > > # Add compiled classes dir to classpath and class_dirs > channels > channels.classpath << @destdir > + channels.class_dirs << @destdir > end > > def build_path > Index: raven/tasks/gem_task.rb > =================================================================== > --- raven/tasks/gem_task.rb (revision 182) > +++ raven/tasks/gem_task.rb (working copy) > @@ -22,7 +22,7 @@ > # project/module directory structure. Otherwise, just set the > # artifact and project to any value that > suits you. > class GemTask < Rake::Task > - attr_writer :version, :artifact, :project > + attr_writer :version, :artifact, :group > > def execute > super > @@ -60,6 +60,6 @@ > end > end > > -def gem(args, &block) > +def makegem(args, &block) > Raven::Tasks::GemTask.define_task(args, &block) > end > \ No newline at end of file > Index: raven/java_tasks.rb > =================================================================== > --- raven/java_tasks.rb (revision 182) > +++ raven/java_tasks.rb (working copy) > @@ -19,11 +19,16 @@ > require 'set' > require 'raven/tasks/matching_task' > > +# Exclude editor tempfiles > +def excluded_task(task_filename) > + task_filename[0,1] == '.' || task_filename[-1,1] == '~' > +end > + > # Load all files in the "raven/tasks" directory on the load path > ($:).each do |base| > if FileTest.directory ?(base + '/raven/tasks') > Dir.foreach(base + '/raven/tasks') do |task| > - require 'raven/tasks/' + task unless task[0,1] == '.' > + require 'raven/tasks/' + task unless excluded_task(task) > end > end > end > > _______________________________________________ > Raven-devel mailing list > Raven-devel at rubyforge.org > http://rubyforge.org/mailman/listinfo/raven-devel > > > _______________________________________________ > Raven-devel mailing list > Raven-devel at rubyforge.org > http://rubyforge.org/mailman/listinfo/raven-devel -------------- next part -------------- An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070401/c5da96cc/attachment.html From matthieu.riou at gmail.com Mon Apr 2 23:37:22 2007 From: matthieu.riou at gmail.com (Matthieu Riou) Date: Mon, 2 Apr 2007 20:37:22 -0700 Subject: [Raven-devel] Quick Q on Raven deps In-Reply-To: <4611142E.4030708@herzumsoftware.com> References: <20070330153618.260FD5240C64@rubyforge.org> <460D3A24.1010602@herzumsoftware.com> <4611142E.4030708@herzumsoftware.com> Message-ID: Posting in raven-devel as well. Actually I thought it was from the beginning, just didn't pay attention. Hi Jesse, No real design docs no, there's only my blog post that was sort of introducing what I intended to do. But I can get a bit more into the details. - The first step of having transitive dependencies working is to be able to build gems with the dependencies imported from Maven poms. This way you can rely on gem dependency mechanism. I've started on this and most gems are now built properly. There may be a few bugs left as I didn't check many gems but it looked okay. You can have a look at repo_builder.rb, you'll see how gems are built and how the Maven poms are parsed. - Once you have the right gems, installing the gems it depends on is quite easy as RubyGems does the job for you. However a couple of hooks are needed to see what gets installed by ruby gems and to have an idea of what gets pulled by what. I've added these hooks in gem_init.rb by redefining some of the methods of the RemoteInstaller. It's far from ideal yet but that's a start. Then the same hooks can be used to toggle the transitive dependencies resolution on or off and to plug some sort of conflict resolution policies. I've just committed these modifications so you can check them out. This is a work in progress anyway so don't hesitate to ask questions, change things and see how it goes. By using this with your project you'll probably find several glitches and necessary adjustments. And I'll try to help you best I can in the process. Cheers, Matthieu On 4/2/07, jbouwman wrote: > > Hi Matthieu, > > Yes, I'd be interested in trying to help here, as this feature would be > important for a project I'm working on. I bet it would be a good > candidate, since it's a large but reasonably well segmented component > system--where, for instance, there's component that uses Hibernate, > another that uses lots of quartz and calendars, etc. Do you have any > recommended starting points or design docs floating around? > > Best Regards > Jesse Bouwman > > Matthieu Riou wrote: > > Hi Jesse, > > > > I've started working on transitive dependencies but I've been pulled a > > bit > > to other stuff and haven't been able to work on it as much as I would > > like. > > But I'll have to get back to it soon. Would you be interested in > > helping out > > and contributing on this subject? > > > > Thanks, > > Matthieu > > > > On 3/30/07, jbouwman wrote: > >> > >> Super!--thanks for answering my questions so quickly--I'll build jruby. > >> To that -- I see a vendor dir in SVN--are you periodically syncing > jruby > >> sources? I'll peek at the devel list. > >> > >> I read a blog posting of yours on transitive dependencies, curious to > >> know if you (or someone else) has been actively pursuing this. Ivy's > >> strategy is ..I think I'm getting this right.. to allow you to select a > >> strategy to converge the versions of common root dependencies > >> (xerces...), either by interpreting version sequences, or by pinning > >> versions. So you provide the minimum amount of information required to > >> produce an unambiguous, conflict-free closure. Something like that > >> anyways. > >> > >> Thanks again, > >> > >> Jesse > >> > >> > >> > >> Matthieu Riou wrote: > >> > Hi Jesse, > >> > > >> > Raven's trunk is the future Raven 2.0 and yes, it requires JRuby. > >> > However I > >> > wouldn't recommend it for direct usage quite yet, it's still under > >> > development and could break anytime. > >> > > >> > On the other hand the 1.x branch is more stable and doesn't require > >> > JRuby. > >> > You can just checkout: > >> > > >> > svn://rubyforge.org/var/svn/raven/branches/v1.x > >> > > >> > There's also a gem for easy install but I guess you're looking for > the > >> > source branch to hack around. Just try to contribute back your hacks > >> > if some > >> > of them can be useful to other ;) > >> > > >> > Cheers, > >> > Matthieu > >> > > >> > On 3/30/07, Jesse Bouwman wrote: > >> >> > >> >> Hi there, > >> >> > >> >> > >> >> > >> >> Been using raven 1.2.2 -- love it, thanks very much. > >> >> > >> >> > >> >> > >> >> Does raven's 'trunk' require running under jruby? I see the first > >> >> require > >> >> line is 'java'. Is trunk where I should be hacking around? > >> >> > >> >> > >> >> > >> >> Been using ant+ivy for build mgmt, very much would like to use > >> something > >> >> like raven as a replacement. Accidentally looked at maven again > last > >> >> week... truly appalling. > >> >> > >> >> > >> >> > >> >> Best regards, > >> >> > >> >> > >> >> > >> >> Jesse > >> >> > >> > > >> > >> > > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070402/98eb9422/attachment.html From nobody at rubyforge.org Tue Apr 3 00:01:59 2007 From: nobody at rubyforge.org (nobody at rubyforge.org) Date: Tue, 3 Apr 2007 00:01:59 -0400 (EDT) Subject: [Raven-devel] [185] trunk/lib/raven: Transitive gems generation and a first step to transitive installs with RubyGems. Message-ID: <20070403040159.CA0385240CD6@rubyforge.org> An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070403/01b0c79f/attachment-0001.html From jbouwman at herzumsoftware.com Tue Apr 3 18:07:13 2007 From: jbouwman at herzumsoftware.com (jbouwman) Date: Tue, 03 Apr 2007 17:07:13 -0500 Subject: [Raven-devel] Some renaming [was: Re: Quick Q on Raven deps] In-Reply-To: References: <20070330153618.260FD5240C64@rubyforge.org> <460D3A24.1010602@herzumsoftware.com> <4611142E.4030708@herzumsoftware.com> Message-ID: <4612D011.60401@herzumsoftware.com> Hi Matthieu, A few initial questions on naming: 1. Could we perhaps rename the tests in raven/trunk/test to test_*.rb, in line with rake's convention? Trying to keep in touch with what I might be breaking. 2. Change maven_crap.rb to maven_repo (and eventually maven_pom.rb) to keep clear the types of maven objects that we interact with. Sort of a different question is, do you expect the 1...1 jar to gem mapping to be stable? More specifically, I'm curious about what the addition of PLATFORM=java type to the Gem package implies, other than constituent jars appearing under 'ext.' I certainly realize that this might not be settled yet. I'm thinking of 'wars' and 'ears' and other complex 'payload types' that might appear Cheers, Jesse Bouwman Matthieu Riou wrote: > Posting in raven-devel as well. Actually I thought it was from the > beginning, just didn't pay attention. From matthieu.riou at gmail.com Tue Apr 3 19:48:03 2007 From: matthieu.riou at gmail.com (Matthieu Riou) Date: Tue, 3 Apr 2007 16:48:03 -0700 Subject: [Raven-devel] Some renaming [was: Re: Quick Q on Raven deps] In-Reply-To: <4612D011.60401@herzumsoftware.com> References: <20070330153618.260FD5240C64@rubyforge.org> <460D3A24.1010602@herzumsoftware.com> <4611142E.4030708@herzumsoftware.com> <4612D011.60401@herzumsoftware.com> Message-ID: > 1. Could we perhaps rename the tests in raven/trunk/test to test_*.rb, > in line with rake's convention? Trying to keep in touch with what I > might be breaking. Wasn't aware of such a convention but if there is, so be it. Matthew any objection? 2. Change maven_crap.rb to maven_repo (and eventually maven_pom.rb) to > keep clear the types of maven objects that we interact with. Yeah, good idea. I've named it that way on a bad day, after a lot of Maven-induced pain but I guess these names would be better. Sort of a different question is, do you expect the 1...1 jar to gem > mapping to be stable? More specifically, I'm curious about what the > addition of PLATFORM=java type to the Gem package implies, other than > constituent jars appearing under 'ext.' I certainly realize that this > might not be settled yet. I'm thinking of 'wars' and 'ears' and other > complex 'payload types' that might appear No I don't really expect it to stay as a 1-1 mapping forever. For now it's the easiest way but if Raven works well enough, it would be much nicer if projects started packaging their gems themselves in a more intelligent manner. For examples projects with multiple jars (i.e. XDoclet, Spring) could include several jars in the same gem. Wars and ears are also good examples. Cheers, Matthieu Cheers, > Jesse Bouwman > > > Matthieu Riou wrote: > > Posting in raven-devel as well. Actually I thought it was from the > > beginning, just didn't pay attention. > > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070403/3a7e3b98/attachment.html From jbouwman at herzumsoftware.com Wed Apr 4 16:19:58 2007 From: jbouwman at herzumsoftware.com (jbouwman) Date: Wed, 04 Apr 2007 15:19:58 -0500 Subject: [Raven-devel] Repository converter updates Message-ID: <4614086E.6070508@herzumsoftware.com> Hello, I've run the M2 repository converter a few times now and it works pretty well, crashing out occasionally on unresolvable entities in poms. My local version now looks at the pom file in order to pick out groupId, artifactId, and so on, instead of deriving these from the filesystem path. Next thing is to pull out the deps. Attached are a patch for 'maven_crap.rb', as well as a new file 'maven_pom.rb' for that. I see a bunch of warnings on the console as the "pom dom" gets built, so I have a feeling I might be failing to do some kind of package initialization? Cheers Jesse Bouwman -------------- next part -------------- A non-text attachment was scrubbed... Name: maven_crap.patch Type: text/x-patch Size: 13533 bytes Desc: not available Url : http://rubyforge.org/pipermail/raven-devel/attachments/20070404/e561b51c/attachment.bin -------------- next part -------------- A non-text attachment was scrubbed... Name: maven_pom.rb Type: application/x-ruby Size: 1327 bytes Desc: not available Url : http://rubyforge.org/pipermail/raven-devel/attachments/20070404/e561b51c/attachment-0001.bin From nobody at rubyforge.org Wed Apr 4 21:33:58 2007 From: nobody at rubyforge.org (nobody at rubyforge.org) Date: Wed, 4 Apr 2007 21:33:58 -0400 (EDT) Subject: [Raven-devel] [186] trunk/lib: Applying patch submitted by Jesse Bouwman. Message-ID: <20070405013358.E25BB5240C21@rubyforge.org> An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070404/62c0c18f/attachment-0001.html From nobody at rubyforge.org Wed Apr 4 21:34:57 2007 From: nobody at rubyforge.org (nobody at rubyforge.org) Date: Wed, 4 Apr 2007 21:34:57 -0400 (EDT) Subject: [Raven-devel] [187] trunk/lib/raven: Renaming maven_crap.rb for a more suitable (well, say descriptive) name. Message-ID: <20070405013457.E10B8524099D@rubyforge.org> An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070404/d59a4690/attachment-0001.html From matthieu.riou at gmail.com Wed Apr 4 21:38:21 2007 From: matthieu.riou at gmail.com (Matthieu Riou) Date: Wed, 4 Apr 2007 18:38:21 -0700 Subject: [Raven-devel] Repository converter updates In-Reply-To: <4614086E.6070508@herzumsoftware.com> References: <4614086E.6070508@herzumsoftware.com> Message-ID: Hi Jesse, I've just applied your patch, thanks! Actually if Matthew is okay with it I'd be more than happy to grant you commit rights. Matthew? The warnings probably come from the poms. Sometimes they're missing or even malformed XML (wtf?!). You just need to make sure the well-formed ones that include dependencies have been correctly processed and that the produced gems have the corresponding dependencies. Cheers, Matthieu On 4/4/07, jbouwman wrote: > > Hello, > > I've run the M2 repository converter a few times now and it works pretty > well, crashing out occasionally on unresolvable entities in poms. My > local version now looks at the pom file in order to pick out groupId, > artifactId, and so on, instead of deriving these from the filesystem > path. Next thing is to pull out the deps. > > Attached are a patch for 'maven_crap.rb', as well as a new file > 'maven_pom.rb' for that. I see a bunch of warnings on the console as > the "pom dom" gets built, so I have a feeling I might be failing to do > some kind of package initialization? > > Cheers > Jesse Bouwman > > _______________________________________________ > Raven-devel mailing list > Raven-devel at rubyforge.org > http://rubyforge.org/mailman/listinfo/raven-devel > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070404/84979512/attachment.html From behmannlist at gmail.com Thu Apr 5 14:47:33 2007 From: behmannlist at gmail.com (Brian Ehmann) Date: Thu, 5 Apr 2007 14:47:33 -0400 Subject: [Raven-devel] Building and Using Raven 2.0 Message-ID: <249F0874-0ACF-4260-9BB6-0094C352BEFC@gmail.com> Hello Everyone, Earlier today I joined the Raven-Devel mailing list so forgive me if this is a duplicate/dumb question. 1) I checked out the source tree and added raven/trunk/vendor/jruby/ bin to my path so that all ruby commands would be run off of this version. (no other version of ruby is on the path) 2) I built (using the included ant build file) jruby and installed the gems rake and antwrap 3) Next I built the raven gem with the command "rake gem" and installed the resulting raven-2.0.gem file into jruby (with "gem install raven-2.0.gem") 4) Finally I attempted to compile a simple project that compiled properly with raven-1.2.2 however I am receiving the following exception: ========= start error =========== % rake compile (in /Users/behmann/src/opensource/river) /Users/behmann/src/opensource/raven/trunk/vendor/jruby/lib/ruby/gems/ 1.8/gems/raven-2.0/lib/raven/gem_init.rb:107:Warning: require_gem is obsolete. Use gem instead. Exception in thread "main" java.lang.NoClassDefFoundError: org/apache/ tools/ant/launch/AntMain ========= end error =========== Did I miss a step or should I download and add the jar which contains this class to my classpath? I would very much like to help port functionality from ANT to Raven but I'm confused as to why I can not create a functional gem. Thanks for your help, - Brian Ehmann From jbouwman at herzumsoftware.com Thu Apr 5 15:10:49 2007 From: jbouwman at herzumsoftware.com (jbouwman) Date: Thu, 05 Apr 2007 14:10:49 -0500 Subject: [Raven-devel] Building and Using Raven 2.0 In-Reply-To: <249F0874-0ACF-4260-9BB6-0094C352BEFC@gmail.com> References: <249F0874-0ACF-4260-9BB6-0094C352BEFC@gmail.com> Message-ID: <461549B9.6030503@herzumsoftware.com> Hi Brian, I've found it useful to put Ant into the CLASSPATH environment variable, something like: LOCAL_LIB=/usr/local/lib/java COMPILER_CP=$LOCAL_LIB/jdk1.5.0/lib/tools.jar ANT_CP=$LOCAL_LIB/apache-ant-1.6.5/lib/ant.jar:$LOCAL_LIB/apache-ant-1.6.5/lib/ant-launcher.jar export CLASSPATH=$COMPILER_CP:$ANT_CP This seem to do it. I'm quite new to jruby, so I inferred this from looking at trunk/test/run. Some of the load path semantics are still a little hazy...hope this helps? Regards, Jesse Brian Ehmann wrote: > Hello Everyone, > > Earlier today I joined the Raven-Devel mailing list so forgive me if > this is a duplicate/dumb question. > > 1) I checked out the source tree and added raven/trunk/vendor/jruby/ > bin to my path so that all ruby commands would be run off of this > version. (no other version of ruby is on the path) > 2) I built (using the included ant build file) jruby and installed > the gems rake and antwrap > 3) Next I built the raven gem with the command "rake gem" and > installed the resulting raven-2.0.gem file into jruby (with "gem > install raven-2.0.gem") > 4) Finally I attempted to compile a simple project that compiled > properly with raven-1.2.2 however I am receiving the following > exception: > > ========= start error =========== > % rake compile > (in /Users/behmann/src/opensource/river) > /Users/behmann/src/opensource/raven/trunk/vendor/jruby/lib/ruby/gems/ > 1.8/gems/raven-2.0/lib/raven/gem_init.rb:107:Warning: require_gem is > obsolete. Use gem instead. > Exception in thread "main" java.lang.NoClassDefFoundError: org/apache/ > tools/ant/launch/AntMain > ========= end error =========== > > Did I miss a step or should I download and add the jar which contains > this class to my classpath? I would very much like to help port > functionality from ANT to Raven but I'm confused as to why I can not > create a functional gem. > > Thanks for your help, > > - Brian Ehmann > _______________________________________________ > Raven-devel mailing list > Raven-devel at rubyforge.org > http://rubyforge.org/mailman/listinfo/raven-devel > From behmannlist at gmail.com Thu Apr 5 16:46:59 2007 From: behmannlist at gmail.com (Brian Ehmann) Date: Thu, 5 Apr 2007 16:46:59 -0400 Subject: [Raven-devel] Building and Using Raven 2.0 In-Reply-To: <461549B9.6030503@herzumsoftware.com> References: <249F0874-0ACF-4260-9BB6-0094C352BEFC@gmail.com> <461549B9.6030503@herzumsoftware.com> Message-ID: <8BABEBF9-8E28-4976-88E9-790D3FAB04AD@gmail.com> Thanks Jesse, that was it. Ok, another noob question. (forgive me) Way the hell back in revision 49 the libs=(l) method was removed from the DependencyTask class in deps_tasks.rb. As a result, the following, although documented in the rdoc, fails with an undefined method exception for libs= dependency 'compile_deps' do |t| t.libs = Dir.glob('lib/**/*.jar') end If the libs= method is returned everything works as expected. Was this method removal intended? Is the "gem" way of using jars expected to be the only way of adding dependencies? - Brian On Apr 5, 2007, at 3:10 PM, jbouwman wrote: > Hi Brian, > > I've found it useful to put Ant into the CLASSPATH environment > variable, > something like: > > LOCAL_LIB=/usr/local/lib/java > COMPILER_CP=$LOCAL_LIB/jdk1.5.0/lib/tools.jar > ANT_CP=$LOCAL_LIB/apache-ant-1.6.5/lib/ant.jar:$LOCAL_LIB/ > apache-ant-1.6.5/lib/ant-launcher.jar > export CLASSPATH=$COMPILER_CP:$ANT_CP > > > This seem to do it. I'm quite new to jruby, so I inferred this from > looking at trunk/test/run. Some of the load path semantics are still a > little hazy...hope this helps? > > Regards, > Jesse > > Brian Ehmann wrote: >> Hello Everyone, >> >> Earlier today I joined the Raven-Devel mailing list so forgive me if >> this is a duplicate/dumb question. >> >> 1) I checked out the source tree and added raven/trunk/vendor/jruby/ >> bin to my path so that all ruby commands would be run off of this >> version. (no other version of ruby is on the path) >> 2) I built (using the included ant build file) jruby and installed >> the gems rake and antwrap >> 3) Next I built the raven gem with the command "rake gem" and >> installed the resulting raven-2.0.gem file into jruby (with "gem >> install raven-2.0.gem") >> 4) Finally I attempted to compile a simple project that compiled >> properly with raven-1.2.2 however I am receiving the following >> exception: >> >> ========= start error =========== >> % rake compile >> (in /Users/behmann/src/opensource/river) >> /Users/behmann/src/opensource/raven/trunk/vendor/jruby/lib/ruby/gems/ >> 1.8/gems/raven-2.0/lib/raven/gem_init.rb:107:Warning: require_gem is >> obsolete. Use gem instead. >> Exception in thread "main" java.lang.NoClassDefFoundError: org/ >> apache/ >> tools/ant/launch/AntMain >> ========= end error =========== >> >> Did I miss a step or should I download and add the jar which contains >> this class to my classpath? I would very much like to help port >> functionality from ANT to Raven but I'm confused as to why I can not >> create a functional gem. >> >> Thanks for your help, >> >> - Brian Ehmann >> _______________________________________________ >> Raven-devel mailing list >> Raven-devel at rubyforge.org >> http://rubyforge.org/mailman/listinfo/raven-devel >> > From jbouwman at herzumsoftware.com Thu Apr 5 17:01:02 2007 From: jbouwman at herzumsoftware.com (Jesse Bouwman) Date: Thu, 05 Apr 2007 16:01:02 -0500 Subject: [Raven-devel] Building and Using Raven 2.0 In-Reply-To: <8BABEBF9-8E28-4976-88E9-790D3FAB04AD@gmail.com> References: <249F0874-0ACF-4260-9BB6-0094C352BEFC@gmail.com> <461549B9.6030503@herzumsoftware.com> <8BABEBF9-8E28-4976-88E9-790D3FAB04AD@gmail.com> Message-ID: <4615638E.1070109@herzumsoftware.com> Hi Brian, I don't know enough about that code to be sure, but I suspect that the method removal was unintentional, so I suggest you restore it and send the patch to this list I have been working on dependencies for, oh, about a day now, and I'm inclined to think that they should be addable both locally and repository resolution. I'm doing a few things with maven interaction right now (import, transitivity), and have been on this list for about a week, but I'll do the best I can with any other questions you have? Regards Jesse Bouwman Brian Ehmann wrote: > Thanks Jesse, that was it. > > Ok, another noob question. (forgive me) > > Way the hell back in revision 49 the libs=(l) method was removed from > the DependencyTask class in deps_tasks.rb. As a result, the > following, although documented in the rdoc, fails with an undefined > method exception for libs= > > dependency 'compile_deps' do |t| > t.libs = Dir.glob('lib/**/*.jar') > end > > If the libs= method is returned everything works as expected. Was > this method removal intended? Is the "gem" way of using jars > expected to be the only way of adding dependencies? > > - Brian > > On Apr 5, 2007, at 3:10 PM, jbouwman wrote: > > >> Hi Brian, >> >> I've found it useful to put Ant into the CLASSPATH environment >> variable, >> something like: >> >> LOCAL_LIB=/usr/local/lib/java >> COMPILER_CP=$LOCAL_LIB/jdk1.5.0/lib/tools.jar >> ANT_CP=$LOCAL_LIB/apache-ant-1.6.5/lib/ant.jar:$LOCAL_LIB/ >> apache-ant-1.6.5/lib/ant-launcher.jar >> export CLASSPATH=$COMPILER_CP:$ANT_CP >> >> >> This seem to do it. I'm quite new to jruby, so I inferred this from >> looking at trunk/test/run. Some of the load path semantics are still a >> little hazy...hope this helps? >> >> Regards, >> Jesse >> >> Brian Ehmann wrote: >> >>> Hello Everyone, >>> >>> Earlier today I joined the Raven-Devel mailing list so forgive me if >>> this is a duplicate/dumb question. >>> >>> 1) I checked out the source tree and added raven/trunk/vendor/jruby/ >>> bin to my path so that all ruby commands would be run off of this >>> version. (no other version of ruby is on the path) >>> 2) I built (using the included ant build file) jruby and installed >>> the gems rake and antwrap >>> 3) Next I built the raven gem with the command "rake gem" and >>> installed the resulting raven-2.0.gem file into jruby (with "gem >>> install raven-2.0.gem") >>> 4) Finally I attempted to compile a simple project that compiled >>> properly with raven-1.2.2 however I am receiving the following >>> exception: >>> >>> ========= start error =========== >>> % rake compile >>> (in /Users/behmann/src/opensource/river) >>> /Users/behmann/src/opensource/raven/trunk/vendor/jruby/lib/ruby/gems/ >>> 1.8/gems/raven-2.0/lib/raven/gem_init.rb:107:Warning: require_gem is >>> obsolete. Use gem instead. >>> Exception in thread "main" java.lang.NoClassDefFoundError: org/ >>> apache/ >>> tools/ant/launch/AntMain >>> ========= end error =========== >>> >>> Did I miss a step or should I download and add the jar which contains >>> this class to my classpath? I would very much like to help port >>> functionality from ANT to Raven but I'm confused as to why I can not >>> create a functional gem. >>> >>> Thanks for your help, >>> >>> - Brian Ehmann >>> _______________________________________________ >>> Raven-devel mailing list >>> Raven-devel at rubyforge.org >>> http://rubyforge.org/mailman/listinfo/raven-devel >>> >>> > > _______________________________________________ > Raven-devel mailing list > Raven-devel at rubyforge.org > http://rubyforge.org/mailman/listinfo/raven-devel > From matthieu.riou at gmail.com Thu Apr 5 17:03:10 2007 From: matthieu.riou at gmail.com (Matthieu Riou) Date: Thu, 5 Apr 2007 14:03:10 -0700 Subject: [Raven-devel] Building and Using Raven 2.0 In-Reply-To: <8BABEBF9-8E28-4976-88E9-790D3FAB04AD@gmail.com> References: <249F0874-0ACF-4260-9BB6-0094C352BEFC@gmail.com> <461549B9.6030503@herzumsoftware.com> <8BABEBF9-8E28-4976-88E9-790D3FAB04AD@gmail.com> Message-ID: Hi Brian, Actually the libs method has been changed to return an array, just like for deps, instead of being a direct accessor. This gives you more freedom on the way you can add libraries. For example you can do any of these: t.libs << 'lib/ant.jar' t.libs << ['lib/ant.jar', 'lib/xerces.jar'] (the libs array is flattened) t.libs[1] = 'lib/ant.jar' .... Sorry for the breakage, the 2.0 release is probably going to be a bit disruptive for 1.x users but I'll try to come up with some basic migration instructions at that time. Cheers, Matthieu On 4/5/07, Brian Ehmann wrote: > > Thanks Jesse, that was it. > > Ok, another noob question. (forgive me) > > Way the hell back in revision 49 the libs=(l) method was removed from > the DependencyTask class in deps_tasks.rb. As a result, the > following, although documented in the rdoc, fails with an undefined > method exception for libs= > > dependency 'compile_deps' do |t| > t.libs = Dir.glob('lib/**/*.jar') > end > > If the libs= method is returned everything works as expected. Was > this method removal intended? Is the "gem" way of using jars > expected to be the only way of adding dependencies? > > - Brian > > On Apr 5, 2007, at 3:10 PM, jbouwman wrote: > > > Hi Brian, > > > > I've found it useful to put Ant into the CLASSPATH environment > > variable, > > something like: > > > > LOCAL_LIB=/usr/local/lib/java > > COMPILER_CP=$LOCAL_LIB/jdk1.5.0/lib/tools.jar > > ANT_CP=$LOCAL_LIB/apache-ant-1.6.5/lib/ant.jar:$LOCAL_LIB/ > > apache-ant-1.6.5/lib/ant-launcher.jar > > export CLASSPATH=$COMPILER_CP:$ANT_CP > > > > > > This seem to do it. I'm quite new to jruby, so I inferred this from > > looking at trunk/test/run. Some of the load path semantics are still a > > little hazy...hope this helps? > > > > Regards, > > Jesse > > > > Brian Ehmann wrote: > >> Hello Everyone, > >> > >> Earlier today I joined the Raven-Devel mailing list so forgive me if > >> this is a duplicate/dumb question. > >> > >> 1) I checked out the source tree and added raven/trunk/vendor/jruby/ > >> bin to my path so that all ruby commands would be run off of this > >> version. (no other version of ruby is on the path) > >> 2) I built (using the included ant build file) jruby and installed > >> the gems rake and antwrap > >> 3) Next I built the raven gem with the command "rake gem" and > >> installed the resulting raven-2.0.gem file into jruby (with "gem > >> install raven-2.0.gem") > >> 4) Finally I attempted to compile a simple project that compiled > >> properly with raven-1.2.2 however I am receiving the following > >> exception: > >> > >> ========= start error =========== > >> % rake compile > >> (in /Users/behmann/src/opensource/river) > >> /Users/behmann/src/opensource/raven/trunk/vendor/jruby/lib/ruby/gems/ > >> 1.8/gems/raven-2.0/lib/raven/gem_init.rb:107:Warning: require_gem is > >> obsolete. Use gem instead. > >> Exception in thread "main" java.lang.NoClassDefFoundError: org/ > >> apache/ > >> tools/ant/launch/AntMain > >> ========= end error =========== > >> > >> Did I miss a step or should I download and add the jar which contains > >> this class to my classpath? I would very much like to help port > >> functionality from ANT to Raven but I'm confused as to why I can not > >> create a functional gem. > >> > >> Thanks for your help, > >> > >> - Brian Ehmann > >> _______________________________________________ > >> Raven-devel mailing list > >> Raven-devel at rubyforge.org > >> http://rubyforge.org/mailman/listinfo/raven-devel > >> > > > > _______________________________________________ > Raven-devel mailing list > Raven-devel at rubyforge.org > http://rubyforge.org/mailman/listinfo/raven-devel > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070405/bfbf7769/attachment-0001.html From behmannlist at gmail.com Thu Apr 5 21:41:06 2007 From: behmannlist at gmail.com (Brian Ehmann) Date: Thu, 5 Apr 2007 21:41:06 -0400 Subject: [Raven-devel] PATCH: JavacTask In-Reply-To: References: <249F0874-0ACF-4260-9BB6-0094C352BEFC@gmail.com> <461549B9.6030503@herzumsoftware.com> <8BABEBF9-8E28-4976-88E9-790D3FAB04AD@gmail.com> Message-ID: <4A8EE7FD-F4FD-43D1-814B-8C323291F753@gmail.com> t.libs << [Dir.glob('lib/**/*.jar')] Cool, that works.... Ok, first patch. I'm still figuring out the code, so it's not one of the open issues. Instead it's a "it works like this in ANT" patch. Changes are: 1) added a :dest_dir symbol to attr_accessor in the javac task so the user can specify the name of their compiled output dir 2) changed require_gem to simply gem on line 104 of gem_init.rb so that the annoying "Warning: require_gem is obsolete" message goes away. Hope you like it. - Brian PS. One other thing... When running the test "run" script, it fails on the all_tests.rb "require 'leafcutter.jar'" statement since the leafcutter.jar file is not in the same directory. Once the leafcutter.jar file was relocated to the same test directory the tests ran without issue. Is this the desired resolution? -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: JavacDestDir_patch.txt Url: http://rubyforge.org/pipermail/raven-devel/attachments/20070405/80995fd9/attachment.txt -------------- next part -------------- On Apr 5, 2007, at 5:03 PM, Matthieu Riou wrote: > Hi Brian, > > Actually the libs method has been changed to return an array, just > like for deps, instead of being a direct accessor. This gives you > more freedom on the way you can add libraries. For example you can > do any of these: > > t.libs << 'lib/ant.jar' > t.libs << ['lib/ant.jar', 'lib/xerces.jar'] (the libs array is > flattened) > t.libs[1] = 'lib/ant.jar' > .... > > Sorry for the breakage, the 2.0 release is probably going to be a > bit disruptive for 1.x users but I'll try to come up with some > basic migration instructions at that time. > > Cheers, > Matthieu > From rubyforge at foemmel.com Fri Apr 6 15:22:52 2007 From: rubyforge at foemmel.com (Matthew Foemmel) Date: Fri, 6 Apr 2007 14:22:52 -0500 Subject: [Raven-devel] Some renaming [was: Re: Quick Q on Raven deps] In-Reply-To: References: <20070330153618.260FD5240C64@rubyforge.org> <460D3A24.1010602@herzumsoftware.com> <4611142E.4030708@herzumsoftware.com> <4612D011.60401@herzumsoftware.com> Message-ID: <0D2F2EE7-E29E-4B40-AE83-0032352D0D18@foemmel.com> I'm definitely OK with renaming the tests to make them more standard. On Apr 3, 2007, at 6:48 PM, Matthieu Riou wrote: > > 1. Could we perhaps rename the tests in raven/trunk/test to test_*.rb, > in line with rake's convention? Trying to keep in touch with what I > might be breaking. > > Wasn't aware of such a convention but if there is, so be it. > Matthew any objection? > > 2. Change maven_crap.rb to maven_repo (and eventually maven_pom.rb) to > keep clear the types of maven objects that we interact with. > > Yeah, good idea. I've named it that way on a bad day, after a lot > of Maven-induced pain but I guess these names would be better. > > Sort of a different question is, do you expect the 1...1 jar to gem > mapping to be stable? More specifically, I'm curious about what the > addition of PLATFORM=java type to the Gem package implies, other than > constituent jars appearing under 'ext.' I certainly realize that > this > might not be settled yet. I'm thinking of 'wars' and 'ears' and other > complex 'payload types' that might appear > > No I don't really expect it to stay as a 1-1 mapping forever. For > now it's the easiest way but if Raven works well enough, it would > be much nicer if projects started packaging their gems themselves > in a more intelligent manner. For examples projects with multiple > jars ( i.e. XDoclet, Spring) could include several jars in the same > gem. Wars and ears are also good examples. > > Cheers, > Matthieu > > Cheers, > Jesse Bouwman > > > Matthieu Riou wrote: > > Posting in raven-devel as well. Actually I thought it was from the > > beginning, just didn't pay attention. > > > _______________________________________________ > Raven-devel mailing list > Raven-devel at rubyforge.org > http://rubyforge.org/mailman/listinfo/raven-devel From rubyforge at foemmel.com Fri Apr 6 16:07:44 2007 From: rubyforge at foemmel.com (Matthew Foemmel) Date: Fri, 6 Apr 2007 15:07:44 -0500 Subject: [Raven-devel] Repository converter updates In-Reply-To: References: <4614086E.6070508@herzumsoftware.com> Message-ID: <8F955B59-79F6-40C2-A728-6217610CE704@foemmel.com> +1 on making Jesse a commiter. Welcome! On Apr 4, 2007, at 8:38 PM, Matthieu Riou wrote: > Hi Jesse, > > I've just applied your patch, thanks! Actually if Matthew is okay > with it I'd be more than happy to grant you commit rights. Matthew? > > The warnings probably come from the poms. Sometimes they're missing > or even malformed XML (wtf?!). You just need to make sure the well- > formed ones that include dependencies have been correctly processed > and that the produced gems have the corresponding dependencies. > > Cheers, > Matthieu > > On 4/4/07, jbouwman wrote: > Hello, > > I've run the M2 repository converter a few times now and it works > pretty > well, crashing out occasionally on unresolvable entities in poms. My > local version now looks at the pom file in order to pick out groupId, > artifactId, and so on, instead of deriving these from the filesystem > path. Next thing is to pull out the deps. > > Attached are a patch for 'maven_crap.rb', as well as a new file > 'maven_pom.rb' for that. I see a bunch of warnings on the console as > the "pom dom" gets built, so I have a feeling I might be failing to do > some kind of package initialization? > > Cheers > Jesse Bouwman > > _______________________________________________ > Raven-devel mailing list > Raven-devel at rubyforge.org > http://rubyforge.org/mailman/listinfo/raven-devel > > > > _______________________________________________ > Raven-devel mailing list > Raven-devel at rubyforge.org > http://rubyforge.org/mailman/listinfo/raven-devel -------------- next part -------------- An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070406/b11a4987/attachment.html From matthieu.riou at gmail.com Fri Apr 6 16:17:37 2007 From: matthieu.riou at gmail.com (Matthieu Riou) Date: Fri, 6 Apr 2007 13:17:37 -0700 Subject: [Raven-devel] Repository converter updates In-Reply-To: <8F955B59-79F6-40C2-A728-6217610CE704@foemmel.com> References: <4614086E.6070508@herzumsoftware.com> <8F955B59-79F6-40C2-A728-6217610CE704@foemmel.com> Message-ID: Perfect! Do you have Rubyforge account Jesse, so I can add you on the project? On 4/6/07, Matthew Foemmel wrote: > > +1 on making Jesse a commiter. Welcome! > On Apr 4, 2007, at 8:38 PM, Matthieu Riou wrote: > > Hi Jesse, > > I've just applied your patch, thanks! Actually if Matthew is okay with it > I'd be more than happy to grant you commit rights. Matthew? > > The warnings probably come from the poms. Sometimes they're missing or > even malformed XML (wtf?!). You just need to make sure the well-formed ones > that include dependencies have been correctly processed and that the > produced gems have the corresponding dependencies. > > Cheers, > Matthieu > > On 4/4/07, jbouwman wrote: > > > > Hello, > > > > I've run the M2 repository converter a few times now and it works pretty > > well, crashing out occasionally on unresolvable entities in poms. My > > local version now looks at the pom file in order to pick out groupId, > > artifactId, and so on, instead of deriving these from the filesystem > > path. Next thing is to pull out the deps. > > > > Attached are a patch for 'maven_crap.rb', as well as a new file > > 'maven_pom.rb' for that. I see a bunch of warnings on the console as > > the "pom dom" gets built, so I have a feeling I might be failing to do > > some kind of package initialization? > > > > Cheers > > Jesse Bouwman > > > > _______________________________________________ > > Raven-devel mailing list > > Raven-devel at rubyforge.org > > http://rubyforge.org/mailman/listinfo/raven-devel > > > > > > > _______________________________________________ > Raven-devel mailing list > Raven-devel at rubyforge.org > http://rubyforge.org/mailman/listinfo/raven-devel > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070406/ceb92f76/attachment.html From caleb.powell at gmail.com Fri Apr 6 23:58:12 2007 From: caleb.powell at gmail.com (Caleb Powell) Date: Fri, 6 Apr 2007 23:58:12 -0400 Subject: [Raven-devel] Antwrap Classpath Message-ID: Hi, I was thinking of modifying the Antwrap library to dynamically load the Ant jar files at runtime. My idea is that the AntProject instance can be passed an ant_home parameter (by default, it would use ENV['ANT_HOME']. Something like this; @ant = AntProject.new(:ant_home => '/Users/caleb/apache-ant-1.7.0') This would save people the trouble of having to setup their Classpath before using the library. In the case of Raven, you could include the Ant jar files with the gem and everything would be seamless. Would this cause any problems for anyone on the Raven project? Does it seem worthwhile? Cheers! -- Caleb From nobody at rubyforge.org Sat Apr 7 21:55:11 2007 From: nobody at rubyforge.org (nobody at rubyforge.org) Date: Sat, 7 Apr 2007 21:55:11 -0400 (EDT) Subject: [Raven-devel] [188] trunk: added eclipse project Message-ID: <20070408015511.CA6D45240D82@rubyforge.org> An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070407/d9d5e0db/attachment-0001.html From rubyforge at foemmel.com Sat Apr 7 22:44:31 2007 From: rubyforge at foemmel.com (Matthew Foemmel) Date: Sat, 7 Apr 2007 21:44:31 -0500 Subject: [Raven-devel] Minor 2.0 patches In-Reply-To: References: <460E99D3.3060401@herzumsoftware.com> Message-ID: <0DBD9240-62F2-4C22-8280-CCF6B8B75782@foemmel.com> Actually, should we just scrap the vendor/jruby directory altogether? Everyone would have to download JRuby separately (and define a JRUBY_HOME environment variable or something so we can find it) but I'm using JRuby for a few other projects right now, and it would be nice to have them all point to the same JRuby instance. Since we're not associating Raven with a specific version of JRuby any more, I think this makes sense. Any objections? On Apr 1, 2007, at 3:18 PM, Matthew Foemmel wrote: > Hi Matthieu, > > I think we should switch to svn:external. When we first set up that > vendor brach, JRuby was a lot buggier than it is now, and I was > thinking we'd be trying out a lot more changes locally before > submitting them to the JRuby team. But it's pretty stable now, and > as you say they've been quite responsive in terms of getting > patches applied, so I think just using svn:external makes sense. > > Cheers, > Foemmel > > On Mar 31, 2007, at 2:54 PM, Matthieu Riou wrote: > >> Hi Jesse, >> >> I've just applied your fixes, they all make sense. Thanks a lot! >> >> JRuby is imported as a vendor branch (see the vendor directory >> under to repository root). The goal was to be able to patch the >> sources directly if we needed to fix something, as JRuby is still >> not quite there. However they tend to be pretty responsive and >> apply patches almost immediately so maybe we could transition to >> svn:external. What do you think Matthew? >> >> Cheers, >> Matthieu >> >> On 3/31/07, jbouwman wrote: >> Hello >> >> I've seen a few small issues on trunk, and had to make the following >> changes to get it working for me: >> >> 1. In javac_task.rb, had to push @destdir onto the class_dirs >> channel in >> order for jar task to see it. Is this code changing? The comment >> above >> mentions this channel, so it seemed odd that it was missing. >> >> 2. In gem_task.rb, changed the :project attribute writer to :group, >> which seems to be the code's 'mavenistic' intention. >> >> 3. Also in gem_task, changed the task name from 'gem', since it >> collides >> with a similarly named function in JRuby 0.9.8. Not sure what the >> best >> name would be here. >> >> 4. in java_tasks.rb, change to ignore both ".xyz" and "xyz~" >> filenames >> in raven/task directories on the load path, ir order to ignore emacs >> backup files. >> >> I do think it makes sense to use an svn external to pull in jruby-- >> I was >> surprised not to see that, on looking at the sources for the first >> time. I've attached diff output for the above things. >> >> >> Thanks! >> >> >> Jesse >> >> >> >> Index: raven/tasks/javac_task.rb >> =================================================================== >> --- raven/tasks/javac_task.rb (revision 182) >> +++ raven/tasks/javac_task.rb (working copy) >> @@ -87,6 +87,7 @@ >> >> # Add compiled classes dir to classpath and class_dirs >> channels >> channels.classpath << @destdir >> + channels.class_dirs << @destdir >> end >> >> def build_path >> Index: raven/tasks/gem_task.rb >> =================================================================== >> --- raven/tasks/gem_task.rb (revision 182) >> +++ raven/tasks/gem_task.rb (working copy) >> @@ -22,7 +22,7 @@ >> # project/module directory structure. Otherwise, just set the >> # artifact and project to any value that >> suits you. >> class GemTask < Rake::Task >> - attr_writer :version, :artifact, :project >> + attr_writer :version, :artifact, :group >> >> def execute >> super >> @@ -60,6 +60,6 @@ >> end >> end >> >> -def gem(args, &block) >> +def makegem(args, &block) >> Raven::Tasks::GemTask.define_task(args, &block) >> end >> \ No newline at end of file >> Index: raven/java_tasks.rb >> =================================================================== >> --- raven/java_tasks.rb (revision 182) >> +++ raven/java_tasks.rb (working copy) >> @@ -19,11 +19,16 @@ >> require 'set' >> require 'raven/tasks/matching_task' >> >> +# Exclude editor tempfiles >> +def excluded_task(task_filename) >> + task_filename[0,1] == '.' || task_filename[-1,1] == '~' >> +end >> + >> # Load all files in the "raven/tasks" directory on the load path >> ($:).each do |base| >> if FileTest.directory ?(base + '/raven/tasks') >> Dir.foreach(base + '/raven/tasks') do |task| >> - require 'raven/tasks/' + task unless task[0,1] == '.' >> + require 'raven/tasks/' + task unless excluded_task(task) >> end >> end >> end >> >> _______________________________________________ >> Raven-devel mailing list >> Raven-devel at rubyforge.org >> http://rubyforge.org/mailman/listinfo/raven-devel >> >> >> _______________________________________________ >> Raven-devel mailing list >> Raven-devel at rubyforge.org >> http://rubyforge.org/mailman/listinfo/raven-devel > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070407/3380e7a9/attachment.html From matthieu.riou at gmail.com Sun Apr 8 11:42:24 2007 From: matthieu.riou at gmail.com (Matthieu Riou) Date: Sun, 8 Apr 2007 08:42:24 -0700 Subject: [Raven-devel] Minor 2.0 patches In-Reply-To: <0DBD9240-62F2-4C22-8280-CCF6B8B75782@foemmel.com> References: <460E99D3.3060401@herzumsoftware.com> <0DBD9240-62F2-4C22-8280-CCF6B8B75782@foemmel.com> Message-ID: Nope. On 4/7/07, Matthew Foemmel wrote: > > Actually, should we just scrap the vendor/jruby directory altogether? > Everyone would have to download JRuby separately (and define a JRUBY_HOME > environment variable or something so we can find it) but I'm using JRuby for > a few other projects right now, and it would be nice to have them all point > to the same JRuby instance. Since we're not associating Raven with a > specific version of JRuby any more, I think this makes sense. Any > objections? > > On Apr 1, 2007, at 3:18 PM, Matthew Foemmel wrote: > > Hi Matthieu, > I think we should switch to svn:external. When we first set up that vendor > brach, JRuby was a lot buggier than it is now, and I was thinking we'd be > trying out a lot more changes locally before submitting them to the JRuby > team. But it's pretty stable now, and as you say they've been quite > responsive in terms of getting patches applied, so I think just using > svn:external makes sense. > > Cheers, > Foemmel > On Mar 31, 2007, at 2:54 PM, Matthieu Riou wrote: > > Hi Jesse, > > I've just applied your fixes, they all make sense. Thanks a lot! > > JRuby is imported as a vendor branch (see the vendor directory under to > repository root). The goal was to be able to patch the sources directly if > we needed to fix something, as JRuby is still not quite there. However they > tend to be pretty responsive and apply patches almost immediately so maybe > we could transition to svn:external. What do you think Matthew? > > Cheers, > Matthieu > > On 3/31/07, jbouwman wrote: > > > > Hello > > > > I've seen a few small issues on trunk, and had to make the following > > changes to get it working for me: > > > > 1. In javac_task.rb, had to push @destdir onto the class_dirs channel in > > order for jar task to see it. Is this code changing? The comment above > > mentions this channel, so it seemed odd that it was missing. > > > > 2. In gem_task.rb, changed the :project attribute writer to :group, > > which seems to be the code's 'mavenistic' intention. > > > > 3. Also in gem_task, changed the task name from 'gem', since it collides > > > > with a similarly named function in JRuby 0.9.8. Not sure what the best > > name would be here. > > > > 4. in java_tasks.rb, change to ignore both ".xyz" and "xyz~" filenames > > in raven/task directories on the load path, ir order to ignore emacs > > backup files. > > > > I do think it makes sense to use an svn external to pull in jruby--I was > > surprised not to see that, on looking at the sources for the first > > time. I've attached diff output for the above things. > > > > > > Thanks! > > > > > > Jesse > > > > > > > > Index: raven/tasks/javac_task.rb > > =================================================================== > > --- raven/tasks/javac_task.rb (revision 182) > > +++ raven/tasks/javac_task.rb (working copy) > > @@ -87,6 +87,7 @@ > > > > # Add compiled classes dir to classpath and class_dirs channels > > channels.classpath << @destdir > > + channels.class_dirs << @destdir > > end > > > > def build_path > > Index: raven/tasks/gem_task.rb > > =================================================================== > > --- raven/tasks/gem_task.rb (revision 182) > > +++ raven/tasks/gem_task.rb (working copy) > > @@ -22,7 +22,7 @@ > > # project/module directory structure. Otherwise, just set the > > # artifact and project to any value that suits > > you. > > class GemTask < Rake::Task > > - attr_writer :version, :artifact, :project > > + attr_writer :version, :artifact, :group > > > > def execute > > super > > @@ -60,6 +60,6 @@ > > end > > end > > > > -def gem(args, &block) > > +def makegem(args, &block) > > Raven::Tasks::GemTask.define_task(args, &block) > > end > > \ No newline at end of file > > Index: raven/java_tasks.rb > > =================================================================== > > --- raven/java_tasks.rb (revision 182) > > +++ raven/java_tasks.rb (working copy) > > @@ -19,11 +19,16 @@ > > require 'set' > > require 'raven/tasks/matching_task' > > > > +# Exclude editor tempfiles > > +def excluded_task(task_filename) > > + task_filename[0,1] == '.' || task_filename[-1,1] == '~' > > +end > > + > > # Load all files in the "raven/tasks" directory on the load path > > ($:).each do |base| > > if FileTest.directory ?(base + '/raven/tasks') > > Dir.foreach(base + '/raven/tasks') do |task| > > - require 'raven/tasks/' + task unless task[0,1] == '.' > > + require 'raven/tasks/' + task unless excluded_task(task) > > end > > end > > end > > > > _______________________________________________ > > Raven-devel mailing list > > Raven-devel at rubyforge.org > > http://rubyforge.org/mailman/listinfo/raven-devel > > > > > _______________________________________________ > Raven-devel mailing list > Raven-devel at rubyforge.org > http://rubyforge.org/mailman/listinfo/raven-devel > > > > > _______________________________________________ > Raven-devel mailing list > Raven-devel at rubyforge.org > http://rubyforge.org/mailman/listinfo/raven-devel > > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070408/4d8bb548/attachment.html From twa at post.com Sun Apr 8 14:10:14 2007 From: twa at post.com (Tom Ayerst) Date: Sun, 08 Apr 2007 19:10:14 +0100 Subject: [Raven-devel] Minor 2.0 patches In-Reply-To: <0DBD9240-62F2-4C22-8280-CCF6B8B75782@foemmel.com> References: <460E99D3.3060401@herzumsoftware.com> <0DBD9240-62F2-4C22-8280-CCF6B8B75782@foemmel.com> Message-ID: <46193006.7040608@post.com> Works for me Matthew Foemmel wrote: > Actually, should we just scrap the vendor/jruby directory altogether? > Everyone would have to download JRuby separately (and define a > JRUBY_HOME environment variable or something so we can find it) but > I'm using JRuby for a few other projects right now, and it would be > nice to have them all point to the same JRuby instance. Since we're > not associating Raven with a specific version of JRuby any more, I > think this makes sense. Any objections? > > On Apr 1, 2007, at 3:18 PM, Matthew Foemmel wrote: > >> Hi Matthieu, >> >> I think we should switch to svn:external. When we first set up that >> vendor brach, JRuby was a lot buggier than it is now, and I was >> thinking we'd be trying out a lot more changes locally before >> submitting them to the JRuby team. But it's pretty stable now, and as >> you say they've been quite responsive in terms of getting patches >> applied, so I think just using svn:external makes sense. >> >> Cheers, >> Foemmel >> >> On Mar 31, 2007, at 2:54 PM, Matthieu Riou wrote: >> >>> Hi Jesse, >>> >>> I've just applied your fixes, they all make sense. Thanks a lot! >>> >>> JRuby is imported as a vendor branch (see the vendor directory under >>> to repository root). The goal was to be able to patch the sources >>> directly if we needed to fix something, as JRuby is still not quite >>> there. However they tend to be pretty responsive and apply patches >>> almost immediately so maybe we could transition to svn:external. >>> What do you think Matthew? >>> >>> Cheers, >>> Matthieu >>> >>> On 3/31/07, jbouwman wrote: >>> Hello >>> >>> I've seen a few small issues on trunk, and had to make the following >>> changes to get it working for me: >>> >>> 1. In javac_task.rb, had to push @destdir onto the class_dirs >>> channel in >>> order for jar task to see it. Is this code changing? The comment above >>> mentions this channel, so it seemed odd that it was missing. >>> >>> 2. In gem_task.rb, changed the :project attribute writer to :group, >>> which seems to be the code's 'mavenistic' intention. >>> >>> 3. Also in gem_task, changed the task name from 'gem', since it >>> collides >>> with a similarly named function in JRuby 0.9.8. Not sure what the best >>> name would be here. >>> >>> 4. in java_tasks.rb, change to ignore both ".xyz" and "xyz~" filenames >>> in raven/task directories on the load path, ir order to ignore emacs >>> backup files. >>> >>> I do think it makes sense to use an svn external to pull in jruby--I >>> was >>> surprised not to see that, on looking at the sources for the first >>> time. I've attached diff output for the above things. >>> >>> >>> Thanks! >>> >>> >>> Jesse >>> >>> >>> >>> Index: raven/tasks/javac_task.rb >>> =================================================================== >>> --- raven/tasks/javac_task.rb (revision 182) >>> +++ raven/tasks/javac_task.rb (working copy) >>> @@ -87,6 +87,7 @@ >>> >>> # Add compiled classes dir to classpath and class_dirs >>> channels >>> channels.classpath << @destdir >>> + channels.class_dirs << @destdir >>> end >>> >>> def build_path >>> Index: raven/tasks/gem_task.rb >>> =================================================================== >>> --- raven/tasks/gem_task.rb (revision 182) >>> +++ raven/tasks/gem_task.rb (working copy) >>> @@ -22,7 +22,7 @@ >>> # project/module directory structure. Otherwise, just set the >>> # artifact and project to any value that >>> suits you. >>> class GemTask < Rake::Task >>> - attr_writer :version, :artifact, :project >>> + attr_writer :version, :artifact, :group >>> >>> def execute >>> super >>> @@ -60,6 +60,6 @@ >>> end >>> end >>> >>> -def gem(args, &block) >>> +def makegem(args, &block) >>> Raven::Tasks::GemTask.define_task(args, &block) >>> end >>> \ No newline at end of file >>> Index: raven/java_tasks.rb >>> =================================================================== >>> --- raven/java_tasks.rb (revision 182) >>> +++ raven/java_tasks.rb (working copy) >>> @@ -19,11 +19,16 @@ >>> require 'set' >>> require 'raven/tasks/matching_task' >>> >>> +# Exclude editor tempfiles >>> +def excluded_task(task_filename) >>> + task_filename[0,1] == '.' || task_filename[-1,1] == '~' >>> +end >>> + >>> # Load all files in the "raven/tasks" directory on the load path >>> ($:).each do |base| >>> if FileTest.directory ?(base + '/raven/tasks') >>> Dir.foreach(base + '/raven/tasks') do |task| >>> - require 'raven/tasks/' + task unless task[0,1] == '.' >>> + require 'raven/tasks/' + task unless excluded_task(task) >>> end >>> end >>> end >>> >>> _______________________________________________ >>> Raven-devel mailing list >>> Raven-devel at rubyforge.org >>> http://rubyforge.org/mailman/listinfo/raven-devel >>> >>> >>> _______________________________________________ >>> Raven-devel mailing list >>> Raven-devel at rubyforge.org >>> http://rubyforge.org/mailman/listinfo/raven-devel >> > > > ------------------------------------------------------------------------ > > _______________________________________________ > Raven-devel mailing list > Raven-devel at rubyforge.org > http://rubyforge.org/mailman/listinfo/raven-devel > From nobody at rubyforge.org Sun Apr 8 14:50:19 2007 From: nobody at rubyforge.org (nobody at rubyforge.org) Date: Sun, 8 Apr 2007 14:50:19 -0400 (EDT) Subject: [Raven-devel] [189] trunk/bin: User must now specify a JRUBY_HOME environment variable for the scripts to work. Message-ID: <20070408185019.E2D785240A96@rubyforge.org> An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070408/177164de/attachment.html From nobody at rubyforge.org Sun Apr 8 14:53:46 2007 From: nobody at rubyforge.org (nobody at rubyforge.org) Date: Sun, 8 Apr 2007 14:53:46 -0400 (EDT) Subject: [Raven-devel] [190] trunk/vendor/jruby/: deleted the vendor/jruby directory (use JRUBY_HOME instead) Message-ID: <20070408185346.1FB1B5240A96@rubyforge.org> An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070408/9c5f1615/attachment.html From jbouwman at herzumsoftware.com Mon Apr 9 09:47:31 2007 From: jbouwman at herzumsoftware.com (Jesse Bouwman) Date: Mon, 09 Apr 2007 08:47:31 -0500 Subject: [Raven-devel] Repository converter updates In-Reply-To: References: <4614086E.6070508@herzumsoftware.com> <8F955B59-79F6-40C2-A728-6217610CE704@foemmel.com> Message-ID: <461A43F3.4020607@herzumsoftware.com> Hi, Yes--and thanks!--under the username 'jbouwman'. I have a few fixes to put in (most embarrassingly, packing up the pom rather than the jar) Regards, Jesse Bouwman Matthieu Riou wrote: > Perfect! Do you have Rubyforge account Jesse, so I can add you on the > project? > > On 4/6/07, Matthew Foemmel wrote: >> >> +1 on making Jesse a commiter. Welcome! >> On Apr 4, 2007, at 8:38 PM, Matthieu Riou wrote: >> >> Hi Jesse, >> >> I've just applied your patch, thanks! Actually if Matthew is okay >> with it >> I'd be more than happy to grant you commit rights. Matthew? >> >> The warnings probably come from the poms. Sometimes they're missing or >> even malformed XML (wtf?!). You just need to make sure the >> well-formed ones >> that include dependencies have been correctly processed and that the >> produced gems have the corresponding dependencies. >> >> Cheers, >> Matthieu >> >> On 4/4/07, jbouwman wrote: >> > >> > Hello, >> > >> > I've run the M2 repository converter a few times now and it works >> pretty >> > well, crashing out occasionally on unresolvable entities in poms. My >> > local version now looks at the pom file in order to pick out groupId, >> > artifactId, and so on, instead of deriving these from the filesystem >> > path. Next thing is to pull out the deps. >> > >> > Attached are a patch for 'maven_crap.rb', as well as a new file >> > 'maven_pom.rb' for that. I see a bunch of warnings on the console as >> > the "pom dom" gets built, so I have a feeling I might be failing to do >> > some kind of package initialization? >> > >> > Cheers >> > Jesse Bouwman >> > >> > _______________________________________________ >> > Raven-devel mailing list >> > Raven-devel at rubyforge.org >> > http://rubyforge.org/mailman/listinfo/raven-devel >> > >> > >> > >> _______________________________________________ >> Raven-devel mailing list >> Raven-devel at rubyforge.org >> http://rubyforge.org/mailman/listinfo/raven-devel >> >> >> > From matthieu.riou at gmail.com Mon Apr 9 10:45:40 2007 From: matthieu.riou at gmail.com (Matthieu Riou) Date: Mon, 9 Apr 2007 07:45:40 -0700 Subject: [Raven-devel] Repository converter updates In-Reply-To: <461A43F3.4020607@herzumsoftware.com> References: <4614086E.6070508@herzumsoftware.com> <8F955B59-79F6-40C2-A728-6217610CE704@foemmel.com> <461A43F3.4020607@herzumsoftware.com> Message-ID: You should be able to commit now. Welcome! On 4/9/07, Jesse Bouwman wrote: > > Hi, > > Yes--and thanks!--under the username 'jbouwman'. I have a few fixes to > put in (most embarrassingly, packing up the pom rather than the jar) > > Regards, > Jesse Bouwman > > Matthieu Riou wrote: > > Perfect! Do you have Rubyforge account Jesse, so I can add you on the > > project? > > > > On 4/6/07, Matthew Foemmel wrote: > >> > >> +1 on making Jesse a commiter. Welcome! > >> On Apr 4, 2007, at 8:38 PM, Matthieu Riou wrote: > >> > >> Hi Jesse, > >> > >> I've just applied your patch, thanks! Actually if Matthew is okay > >> with it > >> I'd be more than happy to grant you commit rights. Matthew? > >> > >> The warnings probably come from the poms. Sometimes they're missing or > >> even malformed XML (wtf?!). You just need to make sure the > >> well-formed ones > >> that include dependencies have been correctly processed and that the > >> produced gems have the corresponding dependencies. > >> > >> Cheers, > >> Matthieu > >> > >> On 4/4/07, jbouwman wrote: > >> > > >> > Hello, > >> > > >> > I've run the M2 repository converter a few times now and it works > >> pretty > >> > well, crashing out occasionally on unresolvable entities in poms. My > >> > local version now looks at the pom file in order to pick out groupId, > >> > artifactId, and so on, instead of deriving these from the filesystem > >> > path. Next thing is to pull out the deps. > >> > > >> > Attached are a patch for 'maven_crap.rb', as well as a new file > >> > 'maven_pom.rb' for that. I see a bunch of warnings on the console as > >> > the "pom dom" gets built, so I have a feeling I might be failing to > do > >> > some kind of package initialization? > >> > > >> > Cheers > >> > Jesse Bouwman > >> > > >> > _______________________________________________ > >> > Raven-devel mailing list > >> > Raven-devel at rubyforge.org > >> > http://rubyforge.org/mailman/listinfo/raven-devel > >> > > >> > > >> > > >> _______________________________________________ > >> Raven-devel mailing list > >> Raven-devel at rubyforge.org > >> http://rubyforge.org/mailman/listinfo/raven-devel > >> > >> > >> > > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070409/b82743b0/attachment.html From nobody at rubyforge.org Mon Apr 9 12:52:54 2007 From: nobody at rubyforge.org (nobody at rubyforge.org) Date: Mon, 9 Apr 2007 12:52:54 -0400 (EDT) Subject: [Raven-devel] [191] trunk/test: Use same path and classpath values as $RAVEN_HOME/bin/raven -- require $JRUBY_HOME Message-ID: <20070409165254.B818352409FB@rubyforge.org> An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070409/55ec7884/attachment.html From rubyforge at foemmel.com Fri Apr 27 17:39:07 2007 From: rubyforge at foemmel.com (Matthew Foemmel) Date: Fri, 27 Apr 2007 16:39:07 -0500 Subject: [Raven-devel] rake-less prototype Message-ID: Hi all, I came up with a little hack the other day, and I'd like to get people's feedback on it. I've been growing a bit frustrated trying to get build channels and multi-project stuff all working cleanly with Rake (which has required a fair bit of monkey-patching) so out of curiousity I decided to see what I could come up with given a "clean slate" i.e. without using any Rake code. I'm not suggesting Raven should switch to this model, but I kind of like the way it turned out, and would be interested in hearing if anyone thinks there are ideas here worth pursuing. I've attached a file with some sample code - it doesn't rely on any existing code (it doesn't do much, in fact) so you should be able to just download and run it with vanilla ruby. Here's an example of what a build script looks like with the prototype: MainCompile = Javac.new { @classpath << "foo.jar" } TestCompile = Javac.new(MainCompile, JUnit::Classpath) { @srcdir = "target/test/java" } Test = JUnit::Run.new(TestCompile) The main differences between the prototype and the current raven2 code are: * Unlike Rake, there is no conceptual difference between a task *instance* and a task *class* - you can call "new" on any task to create an extension of that task * Tasks are objects that the build script must keep track of (usually by storing them in constants e.g. "MainCompile" above) unlike Rake where they are kept in a hashtable and looked up by name * The @ symbol can be used to define build attributes (I've been calling these build "channels" before, but given the new syntax i think "attribute" is more appropriate) Build attributes are handled as follows: * A task can extend any other task (by calling "new" on it) in which case build attributes are automatically passed from the child to the parent * A task can depend on other tasks, in which case the build attributes for the latter are merged together and passed into the former (the same way that Raven works now) Here's a quick walkthough: === Creating Tasks === The obligatory hello world script looks like this: Hello = Task.new { log_info "Hello, World!" } Here we're just creating a new Task object with a single log statement. Calling "build Hello" would print out "Hello, World!" obviously. === Extending Tasks === Now, let's say we wanted to make this task more generic, so that you could optionally pass in the name of the person we want to greet. We could do this like so: Hello = Task.new { @name ||= "World" log_info "Hello, #{@name}!" } In this case, calling "build Hello" would do the same thing as in the previous example (since we set the default value for "@name" to be "World" in the Hello task). However, we can now also define a task to greet "Homer": GreetHomer = Hello.new { @name = "Homer" } Here we're extending the Hello task, by calling "new" on it and passing in a block. (Hello isn't a class, it just happens to have a method called "new"). So calling "new" on the Task class returns a new Task object, on which you can call "new" again, and so on. So from the user's point of view there's no real difference between a task class and a task instance. They're all just tasks. When the GreetHomer task is executed, its block is run (where we set the value of @name to be "Homer") then its parent block is run (where we print out the greeting). The build attribute "@name" is automatically passed from the child to the parent through a bit of metaprogramming magic. In a sense this is like inheritance, except we don't need to specify which method we're overloading, and we don't need to explicitly call "super". === Dependencies === Dependencies can be declared by passing other tasks as arguments to the "new" method. For example, let's say we wanted to set up our classpath, then call Javac: Classpath = Task.new { @classpath = ["foo.jar"] } Compile = Javac.new(Classpath) Build attributes are passed up the same way they are currently in Raven (i.e. the @classpath attribute defined in MainClasspath will be visible in MainCompile). Note that I had to scrap the "=>" notation that rake uses to declaring dependencies, since the name of the task is no longer passed into the task constructor. === Anonymous Tasks === Sometimes you want to group tasks together, but don't necessarily want to define a separate top-level task w/associated dependencies for each one. The "call" command lets you run tasks from within other tasks. For example, say we want to compile some files and copy some resources over in a single build target: Compile = Task.new { call Javac.new { @srcdir = "src/main/java" @destdir = "target/main" } call Copy.new { @srcdir = "src/main/resources"; @destdir = "target/main" @include = "**/*.properties" } } === Modules === The use of capitalize names for tasks may seems strange, but I've done that so that they can be treated as module constants, and therefore be referred to from other modules. This takes the place of Rake namespaces. For example: module Jdk Javac = Task.new { ... } Jar = Task.new { ... } Javadoc = Task.new { ... } end module MyApp # Here we use the fully qualified name MainCompile = Jdk::Javac { ... } # Here we include the module and use the short name include Jdk TestCompile = Jdk::Javac(MainCompile) { ... } end === Gems === The nice thing about using modules instead of Rake namespaces is that we can now bundle up sets of tasks (and their associated files) and distribute them as gems. For example, we could wrap all the JDK tasks up in a gem and use it like this: require_gem 'jdk' MainCompile = Jdk::Javac { @srcdir = "mysrc" } For tools like JUnit (or XDoclet, or Antlr, etc) we might want to provide two tasks: one to set up the classpath, and one for the actual tool itself. This would be fairly easy: module JUnit Classpath = Task.new { @classpath = "junit.jar" } Run = Task.new { # execute the junit tests ... } end Then if you wanted to use JUnit from your build script you could do something like this: require_gem 'junit' CompileTests = Javac.new(JUnit::Classpath) { ... } RunTests = JUnit::Run.new { ... } === Code === I've attached some sample code showing how this might work - let me know what you think... -------------- next part -------------- A non-text attachment was scrubbed... Name: build.rb Type: text/x-ruby-script Size: 2065 bytes Desc: not available Url : http://rubyforge.org/pipermail/raven-devel/attachments/20070427/fec98e30/attachment-0001.bin -------------- next part -------------- Cheers, Foemmel From sasha.ov at gmail.com Fri Apr 27 22:36:14 2007 From: sasha.ov at gmail.com (sasha.ov at gmail.com) Date: Fri, 27 Apr 2007 19:36:14 -0700 Subject: [Raven-devel] bump#1: Running on Windows -- path not absolute Message-ID: <4632B31E.1050305@gmail.com> An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070427/6a9d1e3d/attachment.html From matthieu.riou at gmail.com Sat Apr 28 11:40:51 2007 From: matthieu.riou at gmail.com (Matthieu Riou) Date: Sat, 28 Apr 2007 08:40:51 -0700 Subject: [Raven-devel] bump#1: Running on Windows -- path not absolute In-Reply-To: <4632B31E.1050305@gmail.com> References: <4632B31E.1050305@gmail.com> Message-ID: Hi Sasha, Raven is mostly ready to use but starting with the trunk is maybe not a good idea if you want a painless experience. Most functionalities work right now but are are untested on several environments (like Windows) and we're probably going to break things as development goes on. So for now I'd recommend sticking with Raven 1.x until the 2.x branch (the trunk at this moment) is stable enough. Cheers, Matthieu On 4/27/07, sasha.ov at gmail.com wrote: > > Dear all -- > > Tired of continious wrestling with Maven, I decided to see how Raven can > help. > > Here's my first bump: `raven import` error: > > Successfully built RubyGem > Name: xml-apis-xml-apis > Version: 1.0.b2 > File: xml-apis-xml-apis-1.0.b2-java.gem > [root] Failed to construct POM: install directory "C:\\Documents > and Settings\\Sasha/.raven/gems/xml-apis-xml-apis-1.0.b2-java" not > absolute > > Tried simple things like hard-linking "C:/home" to "C:\Documents and > Settings" and passing the former to raven as -m (raven ignores this > parameter :-( ). > > Using jRuby 0.9.9 and trunk Raven r191. > > Also would appreciate your feedback on whether it is worth trying Raven at > this point. My Maven projects are not especially complex but sometimes > contain little twists, like Java code signing or Flash components. > > Thanks in advance, > -- Sasha > > > _______________________________________________ > Raven-devel mailing list > Raven-devel at rubyforge.org > http://rubyforge.org/mailman/listinfo/raven-devel > > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070428/3efef526/attachment.html From sasha.ov at gmail.com Sat Apr 28 12:04:51 2007 From: sasha.ov at gmail.com (sasha.ov at gmail.com) Date: Sat, 28 Apr 2007 09:04:51 -0700 Subject: [Raven-devel] bump#1: Running on Windows -- path not absolute In-Reply-To: References: <4632B31E.1050305@gmail.com> Message-ID: <463370A3.3040808@gmail.com> An HTML attachment was scrubbed... URL: http://rubyforge.org/pipermail/raven-devel/attachments/20070428/caf970ea/attachment.html