On Tue, 2003-06-24 at 13:43, Conor MacNeill wrote: 
> I'd like to kick off a discussion on what needs to be done to get Ant 1.6 to 
> a 
> release. I'm just going to ramble through some random thoughts I have been 
> having in no particular order just to get discussion started. 
> 
> I don't have any fixed timeframe in mind, or anything so concrete at this 
> stage but I think we need to start looking at it. After all, I can only stand 
> seeing the javah bug reported so many times. Funny how often there is one bug 
> that gets a lot of airplay that it indicates a release may be a good idea.
> 
> 1. Code cleanup
> 
> there is a lot of code with HACK, EXPERIMENTAL, etc tags which I think we 
> need 
> to look at and decide what, if anything, needs to be done. I'd like to attack 
> some of this with checkstyle, etc to improve overall code & javadoc quality 
> and style conformance, etc.
+1
Maybe have a checkstyle.xml parallel to patch.xml.
> 2. Import task
> 
> I think this needs more testing and work. I currently have a disabled 
> testcase 
> that shows one problem. I think we have had some discussions on the meaning 
> of basedirs for imported stuff without any clear resolution.
> 
> 3. Antlibs
> 
> I'm not sure where this is up to. AddType has gone in but other bits are 
> still 
> to be done. I think once we have a workable antlib, we need to look at the 
> packaging of the Ant optional tasks. I'd like to have polymorphism as part of 
> all this.
In http://issues.apache.org/bugzilla/show_bug.cgi?id=19897
my current patch (6762) unifies types and tasks. Unless there are
objections, I will commit this and prepare a patch for roles

The roadmap could be:
   1) roles (allowing the typedef definition to be optionally
      restricted - this is required to allow current conditions
      /filters etcs to be defined) (+ extendtype)
   2) polymorphism (using ant-type magic attribute) 
   3) loading of an antlib.xml file/resource.
   4) namespace support

I could reorder these - or combine them.

> 4. xdocs proposal & manual generation
> 
> I'm not sure if this is still at the proposal stage or ready for primetime. I 
> think we probably won't progress unless we agree that this is the way to 
> generate the Ant manual and commit to supporting it as part of the standard 
> build process. If it is the way to go, I'd like to hurry it along.
> 
> Right now the setproxy doco is generated and looks a bit different from the 
> hand carved manual pages. We probably want a consistent look in the manual 
> and perhaps even a similar look  to the rest of the Ant site. Also I want to 
> have some PDF generation going.

+1 the generated pages look nice, however this should not blocking for
   ant 1.6.

> 5. lcp.bat replacement
> 
> Yes, I have this in mind for a while and will get something done soon on this 
> next. This will be necessary for Windows operation with the new jar 
> arrangement.

+1
> 
> 6. New Tasks and new features.
> 
> I've got some things I want to add to <parallel> to support timeouts, errors, 
> etc plus the capability to have daemon threads. I've thought about whether 
> <macrodef> could be done and would be a good idea. It would provide a way of 
> composing tasks into larger tasks. Peter has mentioned a system to provide 
> task default attribute settings for standard tasks. I've thought about an 
> <antschema> as a companion to <antstructure>, potentially supporting the 
> polymorphic stuff.
+1 for <macrodef> if it can be done. The <extendtype> task does 
   allows attributes to be predefined but it would be nice to allow sub
   elements to be done the same way. 
> 
> 7. Bug reduction.
> 
> I'd like to go through the bugs like a dose of salts. I'd like to see all 
> committers getting stuck into the backlog. The things that aren't going to be 
> done or are unlikely in the forseeable future (e.g. 
> http://nagoya.apache.org/bugzilla/show_bug.cgi?id=3807) could go, IMHO. 
> Voting for bugs will help us to prioritise them.

Some of the bugs are due to ant.bat and the classpath list
being too long for cmd/command. It may be nice for ant to
load the jars in $ant.home/lib in java code rather than the
launch script, perhaps as part of Costin's classloader task.
The lcp.bat replacement may negate this problem.

> 
> OK, that's a start. Please let me know your thoughts.
> 
> Conor
> 
> 
> 
> 
> --
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
> 


---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to