Re: Doubts about nested %if in spec file

["Jaco Greeff" <jaco puxedo org>]

On 30 May 2002 14:10:39 -0400, "Edward C. Bailey" <ed@redhat.com> wrote :
> Actually, this conversation has crystallized it for me -- the issue is that
> the current spec file is mish-mash of "tagged" information, "meta-code"
> (macros, conditionals, etc.), and actual code.  Pulling the code out of
> this stew, and formalizing the representation of the information would only
> be to the good, in my book...

Bingo! I think you have the missing link, well, at least in my eyes. Most of
the normal "representation" tags have been done in my XML implementation.
There is still quite a bit of work left on the scripts and the
representation thereof. Why? XML is a natural match for the actual data
representation, but not for the actual scriplets. it has been bothering me,
but you've nailed it.

Currently post can look something like: (You can view the example.spec.xml
file in cvs for a full implementation of all the currently supported tags)

<spec>
...
<post>
<script>/sbin/ldconfig</script>
...
</post>
</spec>

It is not the most natural way to do things. How for instance do we do ifs?
Does this become another hack? Or do we allow for the scriplets to be done
in something that was designed to do just that, scripting? With thought it
should be possible to embed the scripts, it will just need decent thought.

Compare the above to the "natural XML fit' implementations of

<spec name="example" version="1.0" release="1" ...>

or

<files>
<file attr="777" uid="rpm" gid="rpm">/var/lib/whatever</file>
...
</file>

Ummm... I love questions if I don't have direct answers ;)

Greetings,
Jaco