ant-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Steve Loughran <ste...@apache.org>
Subject Re: Python implementation for Ant
Date Wed, 07 Feb 2007 15:26:13 GMT
Hans Schwaebli wrote:
> Steve Loughran ...
>    
>   If I dislike something it has nothing to do with religion. You can't avoid discussions
by simply stating they are religious.
>    
>   I am not anti-XML in general, but I don't think this format makes any sense for a scripting
or programming language. Even the first creator of Ant today thinks that choosing XML for
writing Ant scripts with it was not the best idea.

I know about JDD's opinions, but I also wonder if he'd written a new 
language, whether or not it would have been so successful.

>    
>   I work with Ant since two years so I know about what I write. If I still think that
writing Ant scripts in XML is cumbersome, than it is for me. This is not to be dismissed as
a religious belief but my personal professional experience with Ant and programming and scripting
languages.

sometimes it is pretty fiddly.

>    
>   It would be good that something in the softwar world is official or is at least mainstream,
because bosses attach importance to it.
>    
>   Why not offer writing Ant scripts in Jython and XML? Let developers choose. But now
they have no real choice for Ant scripts, only XML syntax.

Whoever offers it has to take on all maintenance and support costs. Or 
leave it to the community.

>    
>   There are a lot of reasons why the XML syntax is not good for writing scripts in that
way. The tool support for XML doesn't help except for the very basic things like well-formedness.

and separation of elements in XML namespaces. Whatever language you use 
needs some scoping mechanism, and ant1.6+ has gone for xmlns, which, 
while undeniably awful, does give you the ability to mix libraries 
without name clashes, and the new
antlib dynamic library load services.

>   It has no use to argue for deprecating XML for Ant and using Jython instead to write
Ant scripts, if there is no critical mass here.

I dont see the ant team ever deprecating XML. Ant has always an implicit 
Java API, though there are less guarantees of stability in the Java APIs 
than the XML one. That's important: we strive to make build files 
consistent over time, and do the same for tasks, but there's no 
guarantee things wont break at the java level, which implies the same 
for jython, java, jlog, scala, etc.

>    
>   So is there a critical mass for improving the way to write Ant scripts with Jython,
or at least open mindedness for this idea? Thats the first question on which the debate depends.

1. Ant 1.7 has added much more support for extending ant with scripts 
-you have <scriptdef>, <scriptcondition>, <scriptfilter> etc. We can 
improve this a lot; I dont think you can <scriptdef> a resource (just a 
file), so its no good in an antlib.

2. I'm not convinced that having a new language (especially a procedural 
one) is the right solution, not with java behind it.

If you are going to do a python build tool, have all the equivalents of 
it in python too, so you have a pure scripted env. That way you can do 
more than just choreograph java stuff badly, you have a build process 
built on top of python. Look a rake for an example, though that (and 
capistrano) assume that there's a unix box underneath, so call out to 
unix a lot.

If you wanted to bring a jython front end into the apache work then you 
can certainly propose it, it could go through the incubator, the way ivy 
is going...you just need to meet the success criteria for that  (active 
dev/user community primarily). I just dont see the ant team saying "for 
ant1.8 we've decided to abandon XML and go for python"

-steve


---------------------------------------------------------------------
To unsubscribe, e-mail: user-unsubscribe@ant.apache.org
For additional commands, e-mail: user-help@ant.apache.org


Mime
View raw message