[CMake] Documentation strategy

[Brandon Van Every]

On 6/21/07, Philippe Fremy <phil at freehackers.org> wrote:
> Brandon Van Every wrote:
> > My conclusion is
> > that Kitware doesn't have the resources to address these things, and
> > the community has to find a way to do it.
>
> If I understood correctly, kitware does not do any money off CMake. They
> just need CMake to be good and usable, so that they can make money with
> their other products. That makes it difficult for them to indeed commit
> resources to that kind of problem.
>
> It's unfortunate.

I'll let Bill comment on Kitware's business model if he wishes to.  I
won't presume to repeat the specifics of our conversation.  But what
Kitware makes money at is definitely an issue here.

CMake is hardly the only open source project with this problem.
Chicken Scheme is in ferment right now about wiki docs, Texinfo
formats, and other stuff.  They have the additional liability of
feeling like they need to write their solutions in Scheme.

The problems of distributed authorship have been solved at the source
control level.  You've got Darcs, you've got Mercurial, you've got
Arch if you're willing to stay in Unix-land.  Open source needs to
solve this at the documentation level.  I'm sure a solution will come,
I just wonder from what corner and how long from now.  Everyone
homebrewing the wiki <--> docs problem over and over again just
doesn't work.

> You are burying CMake a bit quickly. How many build source tools out
> there are easy to use, can generate Visual Studio projects or Makefiles,
> are fully cross-platform and work well ?

http://www.scons.org
"Built-in support for Microsoft Visual Studio .NET and past Visual
Studio versions, including generation of .dsp, .dsw, .sln and .vcproj
files."

"Cross-platform Builds Made Easy: From Microsoft Visual Studio to
ARMCC Using Ant and XSL" http://www.devx.com/wireless/Article/28696
How long until someone automates such an approach?

Premake does it.
http://premake.sourceforge.net/about
http://premake.sourceforge.net/premake_users
It may not do it as well as CMake for all I know, it may not have the
community infrastructure, but the seed of a competitor is certainly
there.  I first encountered it as part of Crazy Eddie's GUI, which is
used in several leading open source 3D engines.

CMake is doing great today.  Over the next 3 years, however, there is
plenty of time for others to catch up and/or surpass CMake.  Then the
docs are going to matter quite a bit more, as CMake's technical
capabilities won't be unique or necessarily the best.  CMake currently
does things well that other systems aren't doing well.  It isn't going
to stay that way.

I'm glad the cross-compilation support is coming, because that's a
major strategic issue also.

CPack could be a huge value add but I don't know where it's really
going.  This is a documentation problem again.

I've watched the evolution of tools and technologies in various
segments of the computer industry for 14 years.  There used to be
dozens of 3d card upstarts.  There used to be dozens of big name PC
manufacturers.  Now there aren't.  Consolidation happens.  The
survivors are the ones that execute the best, all the way along the
race.


Cheers,
Brandon Van Every