Seth, I was wonder how the new metadata effort was coming. I had put aside my efforts towards a yum gui until the metadata stuff came out. But, iirc remember correctly yum gave a tentative data of end of Nov ;) Now that Christmas break is upon us I'm going to work full time on the gui. So, a few points 1. Any effort towards a yumErrors class? I have done a very quick effort of replacing errorlog calls with try blocks/except blocks. But these make yummain look horrible, the call to .takeaction has around 20 exception to catch ;) 2. How about a a mechanism to pass back data to the gui? This is pretty kludgey but-- I added a gui-object to logger. Whenever logger is called, in addition to writing to the file-object, it calls gui-object which just puts the raw string on a status label. 3. Have you done any testing on the time taken to generate the metadata? I am looking at only including the ability to use one type of repo (currently yum). If the repo is not yum/metatdata, I have channelSelection run yumarch and cache the hdrs. Thus, I don't have to worry about what type of repo I'm looking at. ALl of the is it a cdrom/file/dirtree/yum crud. Just point and go. You were right comps it a pretty good system! I was getting confused because so much misc stuff was getting swept into comps.py in rcp/up2date. I've done a rewrite of [s|r]cp.comps and pulled out all of the package [sorting|selecting|bestarch|test|install|...] stuff and replaced them with calls to yum. Thanks Dave Farning
