[M3devel] CM3 RELENG: suggestion for distribution packages

[Jay]

> problem sorting out library packages and 
> dependicies - same as we have right now. 

We have a problem sorting out dependencies?
I thought the main dilemna was one of "taste" -- too many installable packages vs. too few. Or is it that "taste" and "dependencies" clash?
 
> Solution to this is order detection, and some topological
> sorting voodoo

If you are satisfied with a linear order that works, pkginfo.txt has one.
 
If you want something that allows for concurrency, well, we don't have infrastructure currently to take advantage of it.
 
I believe a non-linearized graph is not difficult to build given what cm3 knows by reading the m3makefiles. Graphs often make my head hurt though.
 
We should probably, in some future release, 1) move package-to-source-directory feature into cm3 2) allow cm3 to build multiple packages concurrently.
 
 - Jay



----------------------------------------
> From: dragisha at m3w.org
> To: wagner at elegosoft.com
> Date: Wed, 27 May 2009 12:36:34 +0200
> CC: m3devel at elegosoft.com
> Subject: Re: [M3devel] CM3 RELENG: suggestion for distribution packages
>
> Long time ago, I was Modula-2 programmer with lots of my own libraries
> doing everything from keyboard input handling and textual windows,
> relational database, to complex data browsing and input and postscript
> reporting... After a while (2-3 yrs of dev) I had a problem sorting out
> library packages and dependicies - same as we have right now.
>
> Solution to this is order detection, and some topological sorting voodoo
> with result grouping... In my (historical) case I've got six groups of
> modules with controlled inter-group dependicies. My math terminology is
> a bit rusty, but partially ordered set of all packages is our input
> data, and output data we get from such a process is partially ordered
> set of groups of packages. Each element of this final set is one of our
> release packages.
>
> min, std, core, .... is going to my nerves since long time ago. I am
> surely not only one involved with same reaction :).
>
>
> On Wed, 2009-05-27 at 08:57 +0200, Olaf Wagner wrote:
>>
>> The classification of packages is somewhat arbitrary, though I tried
>> to make some meaningful distinctions; it is open for discussion.
> --
> Dragiša Durić 
>