On 30 Jun 2003, seth vidal wrote: > On Sun, 2003-06-29 at 13:29, kenneth topp wrote: > > 1) I used to comment out baseurl to have yum skip over a repo, but that > > generates a stacktrace now: > > > > Traceback (most recent call last): > > File "/usr/bin/yum", line 45, in ? > > yummain.main(sys.argv[1:]) > > File "yummain.py", line 106, in main > > File "yummain.py", line 65, in parseCmdArgs > > File "config.py", line 123, in __init__ > > File "config.py", line 247, in _doreplace > > TypeError: expected string or buffer > > > > I should catch the error - but commenting out a baseurl is not enough to > remove the repo entry. it used to be.. what would it do with an entry otherwise? > > 4) allow users to generator .hdr files when there isn't a repo.. If I > > point yum at a directory full of rpms, why can't it just make .hdr files > > and not require any server side setup at all? It seems that both ftp and > > http support retrieving byte ranges, this would be nice as then all > > directories with rpms in it are yum repos.. It seems that any yum setup > > on servers should be optional.. > > it's not sensible and it's expensive - it's not a mode of operation I > find to be interesting - if someone else wants to implement it fine - > I'll look at it - but it's not as trivial as it seems at first glance > and its not terribly useful. > > yum-arch /some/local/path > baseurl=file:///some/local/path > > just as good - I use it for local rpms I want. I have a standing repo > called local that I can throw rpms in, run yum-arch . on them and be > done. as a dir it gets horribly crufty but <shrug> I clean it out once > every month or so. i guess I'll look and see how expensive it is.. For schools and corporations and the like I guess it's not useful, but for me, it turns every directory full of .rpms on the net into a repo, which is quite useful for me. I can point yum right at ftp.redhat.com, or any official mirror, or redhat cd's, or whatever else.. > > > 5) what is the point of keeping .hdr files in /var/cache that are no > > longer on the server? (debugging?, saves d/ling everything again if a repo > > was flakey?) Whatever it is, it shouldn't be the default... > > yum clean old-headers. > > removing them each time doesn't really help all that much especially > considering how little space they use and sometimes you'll find you're > removing and deleting the same headers A LOT if you did it each time. > well after a couple months I got up to 300MB, and cleaning it (just "yum clean") got me down to 60MB. Why do you think I'd be removing the same entries a lot? Perhaps there is a way that we can remove the files don't correspond to files available on the server anymore? I don't see why yum must require maintance. > -sv > thanks for the good tool..
