[CMake] CMake Marketing
Brandon J. Van Every
bvanevery at gmail.com
Wed Dec 14 20:12:56 EST 2005
Chris Volpe ARA/SED wrote:
> Somewhat related: I read recently that the Delta3D folks are switching
> from CMake back to autoconf due to dissatisfaction/frustration with
> CMake.
>
Where are you reading this? Can you provide a link to the post? When I
Google around for info on this, I only find posts on the subject that
are 8+ months old. It seems they replaced CMake with SCons, not
Autoconf. I downloaded the sources and they now have a SCons based
build in them. Apparently their thinking was driven by a need to
support Linux. Previously they were doing mostly Windows and not much
Linux. They thought about Autoconf, probably not realizing how
unworkable it is if you care about VC++. Someone suggested SCons, which
is much saner, so unsurprisingly they went with it. It looks like they
were either using Python anyways, or intending to.
SCons is a near competitor of CMake. When I was shopping for a better
Chicken Scheme build, my short list came down to CMake and SCons. In
fact there's some other guy working on a Chicken SCons tool, but CMake
was blessed for Chicken builds. Here, CMake had the advantage because
it (1) didn't introduce a new language dependency, (2) was purpose built
to replace Autoconf, which we already had. So, mindset and methodology
were similar. This pattern will recur with any language implementation
that relies on GCC for its bootstrap. I suggest these are people to hit
up. Language guys don't want another language to boot their language,
unless it is C.
Studying the strengths and weaknesses of SCons is going to be important
though. It sits in a similar ecological niche and is most likely to eat
CMake's lunch.
> Has anyone talked to them to find out more details?
>
>
>
That's a good idea, although the picture is probably like I outlined above.
The main downsides I've had with CMake while learning it over the past 2
months have been:
- incomplete documentation. This is a big problem. It's only because
I'm strongly conditioned to read archives, use Google, grep source code,
bug people on mailing lists, etc. that I persisted. In other words I'm
an "Open Source Survivalist." Normal people aren't, and won't go to
such elaborate lengths to try something out. The current state of
electronic documentation is not acceptable for mass awareness, i.e.
having to hunt and peck on the wiki to find variable names, and even
then the listings aren't complete.
- bugs. But, these get fixed, and meanwhile there are workarounds. I'm
not worried about bugs so much.
Cheers,
Brandon Van Every
I won't spend more than 1 day configuring 1 thing.
More information about the CMake
mailing list