ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Peter Donald <dona...@apache.org>
Subject Re: [VOTE] The late stuff
Date Thu, 26 Apr 2001 13:53:57 GMT
At 03:05  26/4/01 +0200, Stefan Bodewig wrote:
>If Jose Alberto wants to run the depend task (i.e. not disable it),
>the whole dependency checking will run once for each compilation unit
>instead of once for the complete build.  And this is sloooow.

He can disable it and then manually insert <depends/> task. We shouldn't be
designing ant around exceptions - we should be making it simple for common
case and make it possible for the other cases.

>> I am concerned that we are sacrificing usability here.
>
>Don't think so.

What you don't think I am concerned or you don't think we are sacrificing
usability? ;)

>> The cases where you don't want to use it (ie jikes, multi-stage
>> compiles, other) is by far more rare than when we do want to use it
>> (ie all the other times).
>
>IMHO the opposite is true.  You don't want to run depend unless you
>know you've changed the non-private interface of a class or
>interface.  No need to recompile all classes depending of A if I only
>change the implementation of a method in A.

Ahh - you mean trust the prgrammers are aware of all the implications of
changes they make? Doesn't that go against the grain of every software
engineering aim. 

I think we should be doing common safest case by default but still allow
people who think they can manage better than the tools. Just because
advanced users can use the tool doesn't mean we should make it harder for
the basic users when we don't have to.

Cheers,

Pete

*-----------------------------------------------------*
| "Faced with the choice between changing one's mind, |
| and proving that there is no need to do so - almost |
| everyone gets busy on the proof."                   |
|              - John Kenneth Galbraith               |
*-----------------------------------------------------*


Mime
View raw message