On Thu, Nov 16, 2006 at 09:15:36PM +0100, Thomas Hartwig wrote:
> >The authors of yum (mostly Seth, but I and most of the rest fully
> >support him) have deliberately chosen a design philosophy which
> >encourages clean, safe management. There are very few "override"
> >options and very few ways to work around problems elsewhere (like
> >packaging problems or repo problems). This encourages power users to
> >do it the "right way" and it helps keep novices from getting into
> >trouble.
>
> Yes indeed, this is it - thank you your understanding ;-)
> But what is the main reason for this philosophy?
Well, my last sentence there was intended to be the reason. Beyond
that, I'm not sure what you mean.
> I think of yum as the major software management tool, which even
> should have all features and not try to wire.
That's a fine design philosophy, too. There are many costs to
features, though:
* it takes coder time to make them
* it takes coder time to maintain them (when random guys submits a
patch, will he be around to maintain it, or will others have to)
* it takes more coder time to debug the entire program
* it may be a "dangerous" feature (see above)
* more options make a program more intimidating and confusing
* a larger and more complex codebase makes things slower and bigger
* a larger and more complex codebase makes it harder for authors to
be familiar with the code
These are all reasons to keep only "essential" features in yum proper
and let the crazier stuff go in utils or plugins.
> There are tools on top of yum already with nice graphical interfaces
> for end users. But this might be my very personal thinking.
a) that's relatively new
b) we don't write or maintain them, and so do not want to rely on
them to guide our design choices
-Michael
--
Michael D. Stenner mstenner@xxxxxxxxxxxxxxx
ECE Department and Optical Sciences Center 520-626-1619
University of Arizona ECE 524G
_______________________________________________
Yum mailing list
Yum@xxxxxxxxxxxxxxxxxxxx
https://lists.dulug.duke.edu/mailman/listinfo/yum
