Re: link-compatibility poll: results and feature removal

From: Zbynek Winkler (zwin_at_[hidden])
Date: 2004-03-22 07:14:06

Vladimir Prus wrote:

> If I understand correctly, the approach you've arrived at is to make feature 
> non-optional, and the drawback is that now all targets are built in four 
> variants? In other words, it's good to have features specific to a main 
> target?
> 
> For example:
> 
> feature mode : 1 2 3 4 
> : link-incompatible ignored-unless-explicitly-made-relevant ;
> 
> lib test : test.cpp : <explicitly-relevant-feature>mode 
> : : <explicit-relevant-feature>mode ;
> 
> exe app : app.cpp test ;
> 
> With this setup, 'mode' feature will only affect target which have explicitly 
> specified that they care about 'mode'. What do you think? (Note that this 
> just an idea, syntax details might be different).

So when I build 'app' - will it have mode set? (Note: for me to be of use it 
should and it should be the default value of the feature 'mode') When I issue 
'bjam mode=1 app' I'd like 'app' to be built with mode set to 1 and linked 
with 'test/<mode>1'. In addition when I do 'bjam mode=1' it should build all 
targets but only 'test' and 'app' should have mode set. Is this what you are 
suggesting? If so, then it solves my problem ;-)

Zbynek

-- 
http://zw.matfyz.cz/ http://robotika.cz/
Faculty of Mathematics and Physics, Charles University, Prague, Czech Republic