my little tooth on cmake

[Kuba Ober]

On Sunday 19 March 2006 07:48, Cristian Tibirna wrote:
> On 17 March 2006 10:17, Nicolas Goutte wrote:
> > > I find it a very good thing that we move away from auto* (these always
> > > scared me white) but we will need to put our collective heads into an
> > > acceptable solution for this level of indirection problem. Precompiled
> > > binaries for most common platforms, copy in our cvs... _I_ don't know,
> > > but _we_ will have to.
> >
> > I am not sure to understand your point.
> >
> > Do you mean now (and in short-term), as for a few features cmake CVS is
> > needed or do you mean on long-term?
> >
> > However on long-term, is cmake.org not supposed to have source and/or
> > binary packages?
>
> Well, exactly. What kills the horse is the uncertainty. Build tools we
> use(d) 'till recently are a mainstay 

Your definition of mainstay seems to imply that a tool merely having the same 
name implies that it will work. Each main kde release had different autotools 
version brackets, and quite often there were very few distros that actually 
included those autotools by default. IIRC sometimes even a minor release 
would obsolete autotools that were mainstay on a particular distro that 
somehow I was using at the time (Murphy's laws?). It was a big PITA more 
often than not. autotools != autotools.

I'd much rather have a tool like cmake, that has less historical baggage and 
being way cleaner. Cmake and kde have short history and it's only natural 
that some things have to be ironed out, but it's not like kde 4 will be 
released a month from now either. Myself having used qmake I don't think that 
the 'extra level of indirection' hurts at all. Maintaining makefiles by hand 
is much worse. Having had suffered from trying to get scons + Qt + aegis play 
ball around January '05 (and ultimately failing as too much scons had to be 
broken apart and changed to get it to perform seamlessly), I'm happy about 
the decision to go ahead with cmake, actually.

Cheers, Kuba