ant-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
Subject RE: Ant -> Make? Ick! - Never!
Date Wed, 18 Oct 2000 01:51:29 GMT
On Tue, 17 Oct 2000, Jose  Alberto Fernandez wrote:
> I guess I did not express myself correctly:
>  Test for equality: if="a=b"
>  Test for inequality: unless="a=b"
> the test for inequality is already there, along with the test for equality
> so there is nothing to fear on that regard.

but there is always a little step further someone
wants. Remember no one wanted if and unless attributes
originally and it was only through much "discussion" that
these were allowed.

> With respect to less-than and greater-than, properties are uninterpreted
> strings hence I do not think one can just "add" these operations without a
> way to declare the properties as something other than Strings. That would be
> adding new data-types.

properties are actually interpreted. Since stefans initial
work properties are now data-types. 

> The only thing left would be lexicographical comparison, but that to me is
> such an useless thing unless you are doing a SORT (not the case here) that I
> doubt it will ever come up.

no but people will want integer comparison and perhaps
mebership in set and then ....

> In my opinion, there is a clear line that we can set. In which equality
> comparison is in and all the rest are out.

they use to say that about if/unless attributes and that
clear line was set there. No matter how much you want it
this will always lead to feature creep. It is not so much a
matter of convincing me or any current active members but
the people who set the project up initially. And quite
frankly you have got buckleys ;)

I want certain features (templates/other logic) in some of
my build files but I recognize it will never make ant core. 

> > >>  DH> (although I think it's perfectly reasonable to offer it as an
> > >>  DH> optional task, since optional tasks should be able to 
> > do whatever
> > >>  DH> anyone wants them to)
> > >> 
> > >> We agree here as well - and I don't think Jose Alberto's <case> has
> > >> been ruled out completely, just put into a loop, waiting for the
> > >> extension mechanism to come.
> > >> 
> > >
> > >This is nice to hear. Of course, if I were to convince you 
> > guys that "if"
> > >is just as good. I will be even happier.
> > 
> > I am pretty sure that it would never get through. There are a lot more
> > committers on this project than who are active and would 
> > oppose it once
> > they found out about it. Hence I will say it again - do this 
> > stuff in a
> > frontend and stream new build.xml to Ant. If you want the 
> > functionality
> > badly add it ;)
> > 
> Well, to tell you the truth there is nothing stoping me to just posting
> my <case> task. And if ANT really gets popular, with the APIs for 2.0
> there will be nothing stoping anyone from defining whatever tasks they want
> and offering them on the net. Are the committers planing to play the 
> thought police game?: "Such task is forbidden to be written" I doubt it.

no but they won't be putting it in core. So you are free to
try and compete - it is a healthy thing. But you can not
expect the basic tenents of ant to change because a minority
wants them,

> > personally I favour xslt to do this and everything else been 
> > asked but as
> > long as frontend interface is there anyone can do whatever ;)
> > 
> So how is a frontend suppose to fix this? Do I need to create yet another
> XML variant so that the "frontend" reads properties files and takes action?


> I hope you are not expecting every developer in a project to maintain their 
> own versions of the XSLTStylesheets (which are way more cryptic than ANT
> itself).

umm - look at xslt - it could be basis of build.xml file and
can do all the logic you want and be just as difficult or
simple as you want. 

If you want power and extreme flexability then you are going
to have to do it as it will never make core - no ifs no
buts. So is it better to complain about how your needs are
not served or try and implement a solution ?



| Latrobe University,     | Does the name 'Pavlov' |
| Bundoora, Australia     |    ring a bell ?       |

View raw message