[Rpm-metadata] take 3 (or is it 4) whatever :)
skvidal at phy.duke.edu
Wed Oct 15 05:04:47 UTC 2003
> No ! Do not mess with the namespace, and make it dependant to
> who have run the package. Your suggestion is NOT a appropriate
> namespace use.
I think this is a confusion. My point was to make it easier for parsers
to see what is an rpm and what is a deb, so I could easily toss out the
debs or rpms depending on which I can deal with. Ideally so someone
could write a program to build this format for a repository containing
> You must consider the semantic of the tags to be associated to the
> fully qualified information (namespace name + local tag name)
> i.e. the URI used for the namespace + the tag name
> Waht you suggest implied that the tools would have to drop check
> for the namespace name and hence work on a non-compliant to the
> XML Namespace spec . Do NOT do that.
> The namespace must be unique for the semantic of the tags we define
> For the tags where there is no agreement on semantic, then they are not
> part of the spec and should be implemented in different namespace.
> Do not mess with namespace if you don't understand the model, read
> the spec if needed it's short !
I read this - and all it defines is a nice way of isolating out the
stuff that is not in the global namespace.
as you mentioned on irc:
if we do that then it'd be handy to have a unique package-type
identifier somewhere in the <common> area so if the format is not one
you know about you can stop parsing.
so something like:
<common entries here/>
sorry for the delay in getting back to everyone, been a bit busy.
More information about the Rpm-metadata