On 04/02 08:22 , Florin Iucha wrote:
> On Tue, Apr 02, 2013 at 08:44:01AM -0400, Carl Wilhelm Soderstrom wrote:
> > On 04/01 11:35 , Florin Iucha wrote:
> > > I have seen this guy present his new make replacement at ESC 2011:
> > > 
> > >    http://www.embedded.com/design/programming-languages-and-tools/4228095/Beyond-MakeFles---Building-large-scale-C-projects--ESC-200-
> > > 
> > > ... and it was a disaster.  Read for yourself...
> > 
> > That's a good article, thanks for posting it.
> > However I'm not clear on what this guy's make-replacement was, or how it was
> > a disaster. Could you clarify?
> 
> The name of the tool was 'tup'.

Ah, ok. It wasn't clear to me that 'tup' was *his* tool.

> The main disaster of his idea is that he can establish the build
> correctness without loading the entire dependency graph into the
> memory, just by checking the file stamps.  It sort of works if you
> just build only one binary out of of a bunch of files, but if you
> produce multiple binaries that share an object file, then it breaks
> down.

Ok. I think I see what you mean.

Also, good point about deleting object files.

Perhaps he'll add some 'warts' to tup to deal with those cases. ;)

-- 
Carl Soderstrom
Systems Administrator
Real-Time Enterprises
www.real-time.com