On Fri, Jun 15, 2012 at 12:41:17PM +0800, Daniel Hartwig wrote: > Although relatively simple, it is still nice to have a common debtags > parser rather than each program implementing it's own. Are there > migration plans for this? In principal I do like the idea of libept > being an extension to the apt database.
I agree, it would be good to have a higher level toolset over apt, to be used for higher level UI things like ratings, comments, screenshots and so on, while keeping apt focused on its task of handling dependencies and maintaining the packages in the system. However, the experience with libept made me realise that we need standards more than code. In part it is because of language barriers: software-center cannot use libept, for example, and aptitude cannot use softare-center's python code. In part is because of the diversity of usage models: some tasks will only want to access information about few packages, therefore calling for quick startup times and indexed information, and other tasks will want a global view of the package information, tolerating longer startup times to get a swifter interaction later. I'm not sure many applications actually need special parsers for data in /var/lib/debtags/: libapt has Tag fields, /var/lib/debtags/vocabulary can be parsed with standard Packages file parsers, and apt-xapian-index provides fast boolean tag queries using stable libraries available in most languages. In hindsight, I consider it more of a success to have been able to get to a point where there is high level data available without a need for extra custom code to access it, than it would have been to have had a successful libept :) > Either way, it will be no problem to test and update aptitude's > internal parser to remove libept for the next release. Thanks! Ciao, Enrico -- GPG key: 1024D/797EBFAB 2000-12-05 Enrico Zini <enr...@enricozini.org>
signature.asc
Description: Digital signature
