Hi,

On Tue, 2008-11-04 at 06:58 -0300, Thiago Bastos wrote:
> Hi there,
> 
>         Yes and no, I guess one thing we would loose was the
>         possibility to
>         just add files and have the make system figure out to include
>         them.
>         I don't know if this would still be working with a cmake based
>         system.
>         
> 
> Well, CMake supports file globbing out-of-the-box if that's what you
> meant. I do use it to automatically gather project files.

I would expect this to happen during the cmake run not the make run ?

> 
>         > If you are willing to give it a try, I can help. Once a
>         developer has
>         > ported the build script, it is also very easy for the other
>         developers
>         > to follow.
>         
>         
>         also yes and no, I have seen weired cmake behaviour where
>         after
>         configure some directory entries were set back but the
>         generate output
>         still carried the correct, old directories. So every time you
>         changed
>         something else you had to reset these directories to get the
>         correct
>         makefiles.
> 
> Hmmm CMake does a lot of caching (which is good, since one of the top
> complaints about SCons is that it's too slow).
> All the weird problems I've seen until now were solved by deleting
> "CMakeCache.txt", which forces CMake to ignore & re-cache everything.
> The only usability hindrance about CMake (in comparison with SCons) is
> that you have to run CMake at least once once before you can run
> 'make', since CMake is a makefile generator.
> But that aside, I prefer CMake's usability all the way around. It has
> generators of all sorts, and does a *really* good job generating
> Visual Studio projects (Xcode integration is really good as well, I
> hear).

that for me is the least important bit ;-). My Windows builds are batch
builds anyway just to check if it works. So besides the Visual Studio
Project stuff there still must be a 'Makefile' based way to build it in
a  useful shell that allows secure remote logins (e.g. cygwin via sshd,
the build-in windows shell does not qualify, nor does nmake).  

> 
>         Anyway in general I would first like to see all the dynamic
>         generations
>         (fcd, scanner/parser, osg-config, OSGConfigured) working
>         correctly
>         before starting to seriously consider it, but I guess if KDE
>         is able
>         to get it working with QT (in particular moc) there should be
>         a way
>         to get it done. Just that we should start with them first.
> 
> I bet it'll be much easier than it was with SCons. 

maybe I never tried ;-), but one curious point would be how cmake
handles the fact that some of our generated files must be in the source
tree, not the build tree. 

> 
>         Also I would like to have a similar easy integration  of cross
>         compile
>         targets, e.g. by just flipping a (single) switch. One problem
>         that is
>         still left with this is currently cmake only supports one tool
>         chain for
>         the build tree but for environments like the PPC/Cell
>         combination you
>         need at least two.
> 
> I don't have any experience with cross compile, but CMake is very
> configurable and modular. I guess in the worst scenario you could
> simply have a CMake script that calls another CMake script multiple
> times, with different settings for different tool chains. But that's
> just a guess...

It's trickier than this. You can't even change the toolchain after
the first cmake run, you have to restart in a fresh build directory,
which btw  requires separate build dirs to work

>         The nice thing of cmake though would be that enabling contrib
>         packages
>         would be in a central place (and they finally have a gui for
>         X11 ;-)).
> 
> There are many other nice things. Handling dependencies will get much
> easier. 

Which kind of dependencies, the following from the cmake faq sounds
scary:

'CMake does not preprocess source files while scanning dependencies'

I haven't checked the implications.

Correct and platform independent constistent dependencies and not just
access time based dependencies is the one strong point on the scons
side.


kind regards,
  gerrit



-------------------------------------------------------------------------
This SF.Net email is sponsored by the Moblin Your Move Developer's challenge
Build the coolest Linux based applications with Moblin SDK & win great prizes
Grand prize is a trip for two to an Open Source event anywhere in the world
http://moblin-contest.org/redirect.php?banner_id=100&url=/
_______________________________________________
Opensg-core mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensg-core

Reply via email to