[CMake-Promote] understanding the CMake build

[Brandon J. Van Every]

John Cowan wrote:
> Brandon J. Van Every scripsit:
>
>   
>>> Actually, it does matter to us whether you or Brandon are getting burned
>>> out: it would be a huge job to take over supporting even the CMake build
>>>       
>> >from zero, 
>>
>> Actually I don't agree.  I didn't do the "write obscure code so that I 
>> remain invaluable" thing, I did rather the opposite.  :-)  The CMake 
>> build is currently feature complete, and it's MIT licensed, so you 
>> simply can't start "from zero."  
>>     
>
> It would be understanding and support that would have to start from zero,
> not coding.  And FLOSSfolk have seen repeatedly that unless there's
> a maintainer, people stay away from projects, because of the often
> well-founded suspicion that nobody knows what's going on.
>   

That's true.  I'm currently the only CMake expert, although I do have 
confidence that in time, Felix will pick up what he needs.  I wonder, 
why would anyone else want to understand the CMake build?  Unless they 
wanted to extend it, I can't think of a reason.  Even a core Chicken 
language developer could get by with very little understanding of the 
build, if they were working on an aspect of Chicken that doesn't have 
build-visible effects.

Turning the build into a CMake tutorial, and pushing it at Kitware via 
CMake-Promote, does sound like a good idea.  It would attract people who 
aren't really interested in Chicken, but who want to learn CMake.  It 
would make the build easier to understand.  Other things are needed as 
well though.  CMake online documentation in general is poor.  The 
mailing list is great, you can get all your questions rapidly answered 
there, but documentation is definitely a serious shortcoming.

Not that people are better off with Autotools.  It's better documented, 
but the whole toolchain will drive you nuts.  You still need those 
hunting and pecking skills, because you have to figure out whether what 
you want to do is an Automake thing, an Autoconf thing, a Libtool thing, 
or a GNU Make thing.  I crash learned Automake over the past 3 weeks, 
already knowing what I wanted the build to do from the CMake side, and 
already having an Automake build that worked but wasn't unified in the 
Chicken build tree.  I can't imagine getting motivated to write an 
Automake build from scratch!

Anyways, the reality is I don't have time for tutorializing the CMake 
build right now.  Nor will it be my 1st priority when I do have time.  
1st priority is OpenGL and games.  Actually, those could become vectors 
for both CMake and Chicken promotion.  This is all very future-looking 
though.  Pretty much the same reason why the CMake online docs don't 
improve; everyone's too busy with something more important.


Cheers,
Brandon Van Every

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/cmake-promote/attachments/20060908/d6125693/attachment.html>