On 3/22/07, Kevin Kofler <kevin.kofler@xxxxxxxxx> wrote:
Another problem is that this assumes the available set of packages is static (the database is generated once at build time and then never touched, that's why it is so fast), so in the context of Fedora, it would require regular rebuilds to deal with the new packages which show up regularly in Extras. An alternative approach would be to query the repository metadata directly at runtime using something like repoquery, but that would be a lot slower. It could however be implemented with only a few lines of bash around repoquery (the single line "repoquery --whatprovides /usr/bin/$1 /bin/$1" would be enough for a basic implementation of command-not-found).
new stuff doesnt need to be rushed inside it (they can wait) .. perhaps a yum plugin that will download of filelist.xml.gz once a week (when the user runs yum) and create binary database from filelist.xml.gz .. and considering createrepo metadata not only used by yum .. other rpm distro might be able to benefit too .. a standalone application similar to updatedb for create-binary-db is also a possible alternative ... if the db is too old, just put a warning message for the user to exec the db updater manually... bash-command-not-found might not be useful for 90% of us who has been using linux for a long time .. but for new learners (not newbie whiners) .. it can be useful for them... -- ----------------------------------------------- regards Hikaru ----------------------------------------------- Mohd Izhar Firdaus Bin Ismail Amano Hikaru 天野晃 「あまの ひかる」 Universiti Teknologi PETRONAS ICT 2nd Year 1st Semester mohd.izhar.firdaus@xxxxxxxxx ----------------------------------------------- kagesenshi.87@xxxxxxxxx Blog: http://kagesenshi.blogspot.com -----------------------------------------------
