Alle 21:25, domenica 19 giugno 2005, John Myers ha scritto: > Problems I see: > - We're talking about a _lot_ of data here, if the service is > widely deployed. After importing most of my packages, my database is > 62MB., plus space from my filesystem for my file name->id mapping > (though the largest table is the one with just the filenames and file > IDs) > - The entire point of this project is eerily similar to that of > most spyware programs: automatic collection of data about users' > usage habits. - In order to be useful, I have to get the involvement > of a large portion of the Gentoo user base. > > Feedback I'd like: > - Is this a worthwile project? Should I continue? > - Is it feasible with that much data? > - Is my design anywhere near decent? > - Is the name stupid? :-P
I think that the database can have only the files names: the ebuilds' maintainers can release the list of the file products from each ebuild without the path. The database is smaller (the path makes it a lot bigger...) and the module system helps a lot to update the list. The service can be as web-application or locale and the database can use a diff file to update... These are my 2 cent! I hope that your application works good! I have always this problem too! Luigi -- Public key GPG(0x073A0960) on http://keyserver.linux.it/
pgpBgyoQLiH5i.pgp
Description: PGP signature
