Re: metadata filenames and other issue for keying the data
seth vidal <[email protected]>
| Newsgroups | gmane.linux.rpm.metadata |
|---|---|
| Message-ID | <1067407114.3522.107.camel@binkley> |
> I don't see the two-file system as a serious problem; I'm just looking > at two types of data that I need to suck down somehow. > > So you're proposing that package-manager-specific metadata could be > stored in separate files and indexed by the core file, a la > "release.xml"? it could be done that way - option two would be an pkg-manager-specific namespace inside any of the files. Though I must say the idea of storing miscellaneous pkg-mgr-specific data in a separate, optional file that can be referenced off of the release.xml-ish file does have an allure. > Maybe an example would help. Forgive my denseness; maybe I should just > shut up until you've implemented this and object if I need to then. object whenever, this is all a work in progress :) > Right. But I was referring to a big catch-all miscellaneous metadata > file, namespaced out the wazoo. I agree that separate files are better. interesting point - see above. what you suggested might be easier. > In defense of XLink, while a URI like > "/metagen/packages.cgi?dist=woody&arch=i386#apt_0.5.14" might look a > little odd, it is still more descriptive as a standalone entity than > "11a59f54c0662fca097c29d1356b2f90". And with path virtualization magic, > the former could be made even more readable. it is more descriptive for a human, but for a program they're about the same and the latter requires less parsing. Give me another example of how you'd see XLink being used, maybe a little more verbose example, let me think if I'm understanding it correctly. what I'm thinking of with using the checksum is just a simple reference that is short and easily passed around to discover what it refers to. With the XLink, I arguably have to know the base path to find the file, but with the checksum, it could be possible to query rpmfind (for example), for that string and get the exact package w/o knowing anything else. I'd say it's pretty reasonable that an md5 or an sha checksum is unique enough :) Does that make any sense? -sv