From: Dale A. <da...@ge...> - 2003-10-06 17:31:48
|
Replying to both at once -- peter reilly wrote: > On Monday 06 October 2003 07:54, Stefan Bodewig wrote: > >>On Sun, 05 Oct 2003, Dale Anson <da...@ge...> wrote: >> >>>They both agreed, and added me to the project, >> >>Welcome! > > +1 Thanks! > However it does remove my nice example for the usefullness of > namespaces in ant ;-) > >>Joining forces sounds like a very good idea. Should we announce that >>in some way? >> >>What is the target Ant release for your tasks? We've tried to keep >>the ant-contrib tasks working with 1.5, but some of them require 1.5.2 >>to compile (but not to run) IIRC. I've tried to keep the tasks to Ant 1.4 compatibility, but haven't been diligent in making sure of that. I know that Java 1.4 is required for at least one task (<post> for sure). > > > Now that ant 1.6 is nearly out, the next release of ant-contrib should I > think move to require 1.6. Some of Dale's custom conditions would be a > usefull addition. > > >>>assert -- like java assert >> >>Have you ever really needed that? Just curious. It turns out to be useful in writing build files for unit tests. It can also be useful in making sure that a developer's environment is set up correctly, especially when a build file is shared across platforms. >> >> >>>limit -- sets a timeout on a task >> >>Sounds very useful. >> >>Is this a simple <sequential> with timeout added? Pretty much, although it doesn't extend <sequential>. I've often thought that <timeout> or <maxwait> might be a better name. I wasn't aware that <sequential> had a timeout. Does it? >> >> >>>math -- all java.lang.Math operations are supported >> >>To do what? Setting properties? Interesting. I admit to getting carried away on this one. As I recall, I needed to add and subtract some numbers, Julian dates were involved. Surprisingly, others are using it. >> >> >>>post -- goes with get >> >>Later versions of Ant are supposed to have a <http> task based on >>commons-httpclient. Are you using plain URL code or relying on an >>external library for that. Plain URL code, no external libraries. >> >> >>>call -- like antcall, but no overhead and runs in current project >>>(might be the same as RunTarget?) >> >>Yep, sounds similar. I read through the code, they are not similar, they are identical (!) >> >> >>>All of antelope is licensed with the Apache license, so there are no >>>issues there. >> >>Maybe with the copyright assignements inside the license itself. >>IANAL and all that, but as neither the ant-contrib project nor >>Antelope are legal entities, the copyright assignments don't mean much >>anyway. Yeah, I don't know (and really don't care) about the license stuff. I just copied the blurb from the top of one of the Ant source files and pasted it in. >> >> >>>I was thinking that for the if/else and try/catch/finally to add >>>some features that are in the antelope version to the ant-contrib >>>version >> >>What features would that be? > > > for if: > * Use of attributes in the if task to test properties > * Extra conditions (using a "bool" container). > http://antelope.sourceforge.net/manual/bk03ch05.html and for <trycatch>: * store full stack trace in a property and/or print the full stack trace * break attribute -- If true and a nested task fails, no other nested tasks will execute. If false, all nested tasks will execute regardless of whether a previous task failed. Note that for each failed task, the 'catch' block (if defined) will execute. Default is "true", which means that the <catch> block is executed on the first failure, then the <try> block is exited. An example of where setting to "false" is handy is for grouping unit tests into a single <try>, letting the <catch> log each failure, but continue to run remaining tests. It would look something like this: <trycatch break="false" property="errormsg" stackTraceProperty="st"> <try> <ant antfile="unit_test1.xml" target="test"/> <ant antfile="unit_test2.xml" target="test"/> </try> <catch> <post to="http://some.server.com/testlogservlet"> <prop name="errormsg"/> <prop name="st"/> </post> </catch> </trycatch> > >>>I told Curt I'd follow the list for a while before committing >>>anything. >> >>As far as I'm concerned, you could start committing away. > > +1 > > Peter > > > > ------------------------------------------------------- > This sf.net email is sponsored by:ThinkGeek > Welcome to geek heaven. > http://thinkgeek.com/sf > _______________________________________________ > Ant-contrib-developers mailing list > Ant...@li... > https://lists.sourceforge.net/lists/listinfo/ant-contrib-developers |