Re: my package requires mysql OR MySQL, how do I state that in thespec file?

[Jeff Johnson <n3npq nc rr com>]

Michael A. Peters wrote:

> On Thu, 2003-09-25 at 06:34, Ralf S. Engelschall wrote:
>  
>
> > On Thu, Sep 25, 2003, Ron Arts wrote:
> >
> >    
> >
> > > Subject says it all.
> > > Couldn't find a searchable archive, sorry.
> > >      
> > AFAIK you can't because RPM (still) does not support alternatives in
> > dependencies. But this is a feature I would like to see myself...
> >    
>
> What about a mapping file that is internel to rpm?
> Kind of like how you can install i386 packages on an i686 machine.
>
> Thus - where naming schemes differ, it could be "compat mapped" - and
> not have anything to do with the package. IE -
>
> gtk2 = gtk+2 = { gtk >= 2 }
>
> so that if my package says it needs gtk+2 and you have gtk >= 2 - rpm
> would know they are same thing (because of a file in /var/lib/rpm that
> has the map predefined)

At the highest possible conceptual level, you're mostly correct,
content from the package metadata can be hidden behind a mapping
layer to avoid unfortunate limitations imposed by rpm package format
and content.

In fact, there are several examples of mapping from package content to
install system configuration already in rpm. I shall leave the discovery
as an exercise ;-)

However, a couple of the major problems with mapping as above are:

   1) Who/how/what/when/where for maintaining the mapping. Note no why ;-)

   2) Adding content mapping adds enormous QA complexity. Many depsolvers
   are routinely looking at dependencies from ~500 packages. Presto, 
chango, now
   do the same QA for all possible mapping substitutions. That's a very 
hard problem.

And that's just for starters.

Some sort of general mapping for substitution is needed for rpm 
packages, probably pretty soon too,

73 de Jeff