ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
Subject Re: Death of a mutant ...
Date Thu, 11 Jul 2002 14:33:14 GMT
On 11 Jul 2002, Stefan Bodewig wrote:

> > Improving Ant on the current codebase is not impossible, and it
> > seems many people believe this is the way to go.
> Are you sure you have your numbers straight?  Let me count, Costin and
> myself (and I wouldn't even cast a -1) on the one side and Jon, Erik,
> Peter, Adam, Magesh, Diane and you on the other.

By another count, most people haven't look deep into any proposal. It 
took me a lot of time to understand both ( I started months ago, and I 
still can't say I know them very well ). 

Again - my -1 is on adopting a large chunk of code, mostly unreviewed
( and certainly not the level of review ordinary changes in ant have ).
And I'm -1 on making the core more complicated than it is - and so 
far both proposals do that IMHO, mostly without real gains. 

But of the 2 proposals, Mutant is what I would have probably choosed
if forced to. And I have no problems with adopting a new core or/and
new implementation - if it is clearly better.

What if instead of deleting mutant, you start selecting the most
important features/interfaces and propose them, one by one, to
the main branch ? 

For example in a org.apache.ant.core package. 

Things like the Task interface - we can then spend the time and review it,
decide if  'handleSystemOut' is really needed ( or what it means ). It can
be merged with the equivalent concept in myrmidon, and we can make the 
current Task implement/extend it ( I would prefer it as an abstract 
class - they're many times better for backwared compat ).

Or we can take the SAX2 code and implement it as a ProjectHelper,
eventually after we select a new org.apache.ant.core 
abstraction/interface for this particular operation. I can drop 
the ProjectHelper2 and use your code ( or the code from myrmidon ).

The new core shouldn't be more complex than the current one - so 
it shouldn't have too many or too complex interfaces. And each inteface
needs a full and _VERY_ deep review and discussion - I would ask for
at least 3-4 weeks on each. 
As for JDK1.1 compat - I'm strongly against making the new interfaces 
require 1.2. There is absolutely no justification. I'm ok with using
1.2 in the implementation - since 1.1 implementations will be possible.
I'm willing to volunteer to deal with 1.1 problems in 1.6 - 
eventually by replacing the introspection-based code with the 
tomcat.util.compat package. 

I hope everyone realize how many users ant has - I suspect more people
use ant than JDK1.4 today. There are huge investments in time and 
learning, books, etc. If we really want a new API for future - it better
be _very_ good.


> > BTW, if another committer would like to take over management of the
> > Ant mailing lists, I think that would make sense since I will be
> > taking a break.
> Does anybody know whether it is possible to have more than one
> moderator?
> Thank you, Conor.
> Stefan
> --
> To unsubscribe, e-mail:   <>
> For additional commands, e-mail: <>

To unsubscribe, e-mail:   <>
For additional commands, e-mail: <>

View raw message