At 06:39 15/5/01 -0700, Alan George wrote: >I am sorry, I must be confused,
yes - you seem to be. >I though ant was supposed to be a build tool;) >Why is everybody so hung up on 'if does not make since to do this in XML it must >not be a build problem we need to solve'. I think your focus on XML is mistake. >A build is a work flow not some static data format (i.e. XML). When all you have >is a hammer..... XML has NOTHING to do with the issues at - hand. The same issues were addressed in design of make that doesn't use XML. If you look at any medium size (hand-maintained) project they generally use autoconf+automake or something similar (imake or sets of "template/rule" make files). If at the begining of make the OSes had more memory and similar tools were available - then make would be simple to use. Instead they kept adding features to make so as to combat the lack of auto* tools. In the end very few people can understand the tools and none of them would lay claim to make being simple. We are trying to avoid that (keep things simple) and satisfy the need (for templating et al) via other means. >Also, I can't get around the fact that there are if/unless attributes for >targets. If conditional processing of a build is so bad, why do targets support >this. If you where more consistent, you might not get bozzos off the street like >me saying "give me more". if/unless and conditional processing is not something we are opposed to - all my projects use it. It is per task conditional execution. If we had that - why would we need targets??? If we had that - why would we need ant - we could do the same thing using python and a few modules. 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 | *-----------------------------------------------------*
