Jeff Spaleta wrote:
Are you having trouble finding an official mirrorlist for Core?Not at all. For the test , I'm using two mirrors: mirror.linux.duke.edu and mirror.pop-rs.rnp.br . I've chosen these two because duke.edu is always in synch and por-rs.rnp.br is always missing packages...
For the purposes of testing this idea to detect mirror syncage. II didnt think about restricting the testing to a few directories , so debugging was a bit messy.... I'll try to add some extra functionality to the config files (currently one config option is Nodevel , to disable checking the development dir. I guess I'll do the same for updates/testing and for all the SRPMs directories)
would encourage you to stick initially with just Core and possibly a
special case repository that exists just to test
how this syncage process works instead of trying to test against
multiple repositories in the wild. Is there a prototype cron script
yet that has atleast some functionality? Don't shoot for
perfection...shoot for producing a script that people can test and
give you feedback on and help with.
The script is currently having a problem with the hashes I'm using ... I'll try to change from hashes to arrays (or from a 4 dimension hash to a 2 dimension) and see if it works... The bug is kinda weird.. I can access the data if I use $redhat{"updates"}{"i386"}{"2"} but it doesnt work when I use $redhat{"$directory"}{"$arch"}{"$release"} (and all those vars are set to the correct values)....
Besides that , all functions seem to work perfectly (I have the habit of testing each function separately , so I can fix them faster). Only the function which should read the hashes is weird... Probably I'll have to change it and the one that creates the hashes , but I guess I can do it in one hour or two...
I'll try to fix the code and as soon as it works , I'll post the url for the file it will generate. The source for the current non-working version is currently at www.dcc.ufmg.br/~pmacedo/fedora/test.pl if you want to take a look.
-- Pedro Macedo
