On Tue, Aug 17, 2010 at 12:41:42PM -0400, MK wrote: > "Interface style" and in particular "Software Category" are the > criteria I think most easily/appropriately garnered from debtag's > facet::tag structure. Most of the other data is already available via > apt-cache, so I'd like to use debtags to round out the information; this > way we can add stuff from debian databases without having to leave some > of the wiki criteria blank.
Ok. The only thing I didn't get is what you mean with "leave some of the wiki criteria blank". > > Rather than abandoned I'd call it stable: > > Yes, sorry about that -- I downloaded the CVS onto a fedora system > thinking I could hack a db, then noticed the last "news" on your site > is c. 2006. Anyway, I now have the debtags distro package installed > here on ubuntu and gnusense. Ouch yes, the news need updating. The existing method to update the news required manual intervention and was a heavily bitrotten ruby script. I've now rewritten it with something that can be scheduled from cron, so hopefully it should stay up to date. At some point I'll want to blog about your efforts so it also ends up in the news in the debtags homepage. > Which leads to my first real question(s): I ran "debtags update" and > then tried "debtags show debtags" (or "debtags show vim", et. al.), but > contra the manpage, this just gives exactly the same thing as > "apt-cache show" -- there is no facet::tag data. Yes. In the past, apt-cache show wouldn't show tags, and debtags show would; now apt-cache has caught up with debtags in that respect :) > Without searching for > a specific tag, the only way I seem to be able to get this info is via > "debtags cat". That may actually be sufficient, since the package > names can be paired with an apt-cache dump, but any suggestions or > insight you might have here would be good. Again, I don't need to > search by tag, I need to get tag info by package (I notice "debtags > grep" does not work with the package name). You're probably better off skipping debtags altogether and just getting the raw, reviewed dataset with: svn cat svn://svn.debian.org/debtags/tagdb/tags > > It should be very easy to parse: read one line at a time, use /:\s+/ > > to split packages from tags, and /,\s+/ to split the tag list. > > Okay, so that is the same as the "debtags cat" dump? Yes, the format is the same. In fact, "debtags cat" does little more than outputting /var/lib/debtags/package-tags which has data that is taken from svn://svn.debian.org/debtags/tagdb/tags but is limited to the packages you can install in your system. For that reason I suggest you to get the data from svn: you get all packages regardless of what is your architecture. > > If then you allow people to edit tags in your mediawiki, it'd be > > fantastic if the edits you get could be contributed back. That should > > be also rather easy to arrange, again we'll talk about it if it's > > something you'd like to do and when we'll get there. > > I am developing a php API to allow people to machine query the site > (various groups are interested in mirroring the info in various ways), > so at the very least you should be able to come up with a way to mine > stuff that way. I'll keep this desire in mind as I proceed ;) \o/ A way to machine query the data should be enough. Let's see the details once you have it set up. > > But let's go one step at a time. Is the vocabulary adequate for your > > needs? > > Looking thru that cat dump I think it will not be hard to map this to > the existing "Software categories" (there are 200 or so, taken directly > from the old database) and the interface facet is identical, so this is > starting to look easier than at first I thought. When I come up with a > mapping, maybe I will run it by you for an opinion? Yes, by all means. > Anyway, if you have a few minutes to look at the wiki and lemme know if > anything occurs to you in these regards, it would be much > appreciated. The current format is not carved in stone, and it's the > FSF Board that originally expressed interest in using debtags, so they > (and I) are probably open to ideas vis. a closer correlation to the > facet::tag system (within the peculiar confines of semantic mediawiki). Ok. There are a few quirks interface-wise (UI tags are shown in Prerequisites, in Find the lists only show 2 or 3 lines making it hard to read the options) but I assume those are because the website is not finished yet. You may want to have a look at the works-with:: tags, which I find to be quite useful user-wise. Forget about works-with-format:: though, which should really be superseded by a database of mime types. Have you seen the goplay/goadmin/golearn/... packages? They make use of tags (facets like game::, admin::, biology::, devel::, office::, security::, sound::, web::) to provide task-specific package search interfaces. The result is really rather nice. hardware:: is probably also a facet that could have a use in a package index. These are the first things that come off the top of my mind. Please let me know if they are useful or if you were looking for something else. Ciao, Enrico -- GPG key: 4096R/E7AD5568 2009-05-08 Enrico Zini <[email protected]>
signature.asc
Description: Digital signature
_______________________________________________ Debtags-devel mailing list [email protected] http://lists.alioth.debian.org/mailman/listinfo/debtags-devel
