Package distribution hassles - a proposal.

[Max TenEyck Woodbury]

John H Terpstra wrote:
> 
> On Mon, 4 Mar 2002, Max TenEyck Woodbury wrote:
> 
>> ...
> 
> You may want to check out http://www.tuxfinder.com. This is a
> magic site.

Thanks. I'll look into it...

>> ...
> 
> Looking forward to the next step. How shall we proceed?

I take it that the process and inputs looks correct to you.

Next, I figure out how to specify the inputs, lash up an
example and post the result for your review. While you're
reviewing that, I start to implement the procedure. Be
warned, it will probably be a crude bailing wire and 
chewing gum sort of thing with bash scripts, make files 
and maybe a C program or two to start with. First pass
will NOT do anything except read the spec and announce what
steps it should be taking. You get to check that out
to assure that I haven't pulled any major boners at that
stage.

After that I fix the parts that don't work (i.e. everything).
(If this sounds a little strange, you might be interested to 
know that I learned programming by debugging my father's code.)
A lot of the steps are very straight forward and should fall
into place quickly. Once the simple stuff is in place, that
will need to be checked, while I start putting the more 
complicated pieces in place. You get to review that. At that
point we should have a tool that can build and test at least
one kind of binary distribution. (Almost certainly the current
Red Hat variant.)

After that, we need to try it on older distributions, try adding
patches, and other packagers. Problems will arise and need to
be fixed. If you aren't completely fed up with me and the project
after that, a man page or some other basic documentation would be 
next.

I don't know the details of what happens after that. If you find
it useful, decisions will have to be made about extending it to
a non-rpm environment, better documentation and maintenance. You 
could be so mad at me for wasting your time that you never want 
to see the cussed thing again.

mtew