[SDL] CMake gurus...

Sik the hedgehog sik.the.hedgehog at gmail.com
Sun Mar 24 04:32:19 PDT 2013


Actually, there are snapshots from Mercurial which are just ZIP files
with the source code (OK, other formats available too), no need to
know about Mercurial at all. Those are also easier to build (no stupid
warning regarding not building from the Mercurial repos - the readme
isn't clear on how to work around this, sadly).

2013/3/24, Kai Sterker <kai.sterker at gmail.com>:
> On Sun, Mar 24, 2013 at 4:42 AM, Ryan C. Gordon <icculus at icculus.org> wrote:
>
>> Ideally we want Windows people to find a Visual Studio project that
>> matches
>> their version of Visual Studio, double-click it, and build the library.
>
> Ideally, wouldn't Windows or OSX developers want readymade
> DLLs/Frameworks with an SDL project template for their IDE that lets
> them start work on their own application right away? Why should anyone
> be forced to compile SDL from the sources first?
>
>
>> Right now we're maintaining several .vcproj files for different versions,
>> and almost universally, some subset of them fails to get updated when we
>> change a project setting of add/remove files.
>
> Shouldn't this result in lots of complaints on the mailing list, from
> the people that try to compile SDL using those projects and fail? Have
> seen more requests for binaries than those (although possibly as a
> result of the projects not working?).
>
>
> Trying to think of valid cases why anyone would want to compile his or
> her own version of SDL in the first place. Granted, as long as there
> is no official SDL 2.0 release, there's a need. But from my point of
> view, if somebody is knowledgeable enough to obtain the SDL sources
> from a RCS as obscure as Mercurial, wouldn't they be also able to
> check a README and have CMake generate the desired project file,
> especially if the steps and settings were properly described?
>
> Once SDL 2.0 gets official, outside from actually contributing to SDL
> or making changes outside of contributing, I don't see much need of
> compiling it. Sure, it can be nice to have the source around to check
> how things work exactly, or to debug in case of issues, but for that I
> do not necessarily require a project for my favourite IDE (which btw.
> is Eclipse CDT + gcc or MingW toolchain, and SDL never provided
> projects for that).
>
>
> From a personal point of view, as somebody who has used SDL 1.3 on
> Linux, Windows (w/ MingW) and OSX, I was very happy with the autotools
> build, as it's the same simple steps on each platform. I'd also be
> happy with CMake, as to me it seems speedier than autotools and more
> flexible, although I'd probably prefer to just have it create
> Makefiles.
>
> In my own project, both autotools and CMake are available as the build
> system, and the only reason to not drop autotools is that CMake is (or
> at least was in the past) inferior when it comes to the "make dist"
> step. I never tried generating IDE projects, though, apart from XCode
> and that wouldn't even work out of the box).
>
>
> So in my opinion, switch to CMake, drop the prebuild IDE projects and
> educate the few that may need it how to run CMake. Maybe provide a few
> batch or shell scripts that invoke the cmake command line tool with
> the appropriate switches.
>
> My 2¢,
>
> Kai
> _______________________________________________
> SDL mailing list
> SDL at lists.libsdl.org
> http://lists.libsdl.org/listinfo.cgi/sdl-libsdl.org
>



More information about the SDL mailing list