[cairo] Re[4]: Cairographics on win32

[Mike Emmel]

On 4/17/05, Hans Breuer <hans at breuer.org> wrote:
> 
> Am 14.04.2005 um 16:06 schrieb Mike Emmel:
> 
> > I think a more productive approach is to identify  the failings of
> > autoconf and friends for the Win32 platform and determine if it can be
> > fixed
> If you think fixing auto*tools is a feasible option just go ahead
> and try. I don't.
> 
I don't know to be honest my first impression is to bring in more
build targets besides make I'm pretty sure thats doable. I've looked
some for example at better integraton for java which actually included
cmake and I feel its doable.
The big problem is replacing automake. The big advantage of the
autotools is it looks like you can add a new tool easily if it
understnds the configure.in file.

> > instead of throwing it out.
> Noone is proposing to throw it out. But from my experience with
> auto*tools (on windows) and form the mere existence of other
> cross platform build tools trying one of them looks reasonable
> to me.
> 
I know and I could bring up ant which is quite powerful but I think
like cmake solves a different problem.
> [...]
> > them. Generally I've found "Windows" build tools to break horribly
> > between windows variants so the windows build stuff only works for a
> > few windows enviroments. Autconf can at the minimum tell me what I
> > need to get a correct build enviroment.
> >
> Does not look like a big help if the only way to get it working is
> patching it. I already wasted days of work to get auto*tools running
> for me and only got them somewhat working with the help of some
> experts [1]. Setting up CMake and writing the first version of
> working (for the msvc build) control files was a matter of few hours.
> 
>         Hans
> 
> 

Well autoconf and friends mainly deal with simply setting variables
you don't think that CMake could fit into that framework even loosely.
I've looked at Cmake and its a nice system I just think that the
concept of the auto tools i.e mainly discovering the system and its
build requirements could fit with CMake wich is more a build tool.  It
sounds like your experience was bad but was it the issue's of the tool
is not well supported on windows i.e you had a hard time installing or
the tool itself ?
The gnu toolchain on windows is notorious for its issues and quirks if
you have them working now can't you share the work you did to lessen
the pain for others ?   If the autotools are really deficient then
what are the problems installation basic flaws ??  And cmake may solve
the problem for this one thing for you cairo but since most opensource
projects use autotools your really not solving the major issue. No one
is going to do cmake and autotools to support windows.   In general
for me I like to concept of the autotools I think the implementation
has a long way to go but the heart of the design of a extensible
discovery based build system is correct. So for me despite its short
comings I'm willing to support it because its intrinsic design is
better than ant cmake etc.

Mike