#37 How is recursing dependency trees supposed to work?
Opened 7 years ago by codeblock. Modified 7 years ago

Look at the output of https://apps.fedoraproject.org/mdapi/rawhide/pkg/irssi and you'll notice that the requires section includes things like perl(Exporter). But going to "https://apps.fedoraproject.org/mdapi/rawhide/pkg/perl(Exporter)" produces a 404. The correct URL is https://apps.fedoraproject.org/mdapi/rawhide/pkg/perl-Exporter.

So my questions are:

  • Can we make it a law that any name that mdapi produces in a "name" field will result in a non-404 result? I.e., handle the mapping internally?
  • Otherwise, can we document how the mapping is supposed to work? There's a similar thing with dynamically-added .so requirements.

  • Can we make it a law that any name that mdapi produces in a "name" field will result in a non-404 result? I.e., handle the mapping internally?

So I'm not really sure about this, I really kinda want to keep the logic in mdapi as minimal as possible, to me it should just expose what's in the repo metadata. If there some logic needed, imho it should be in the client, not the server.

For the record (since we already discussed it in IRC), a solution around this might be to use the provides/ endpoints if pkg/ returns a 404:
https://apps.fedoraproject.org/mdapi/rawhide/provides/perl(Exporter)

Login to comment on this ticket.

Metadata