On Apr 1, 2005, at 13:33, seth vidal wrote: > >> sorry to hear that. >> > > I guess i'm just confused what you would have me do. > > Relying on the headers is not going to work. > > I don't think I mentioned headers. I must admit that I still include them in my repositories to allow the user to update yum itself, but that is outside this scope. I use xml in a number of other contexts so I agree that xml is a valid solution to a number of problems. Just - it created one here. As I said: I see a problem. I didn't say: here is what you should do. I don't think your argument about anaconda holds any kind of water. I can wipe the disk any time too. I wouldn't need yum then. The idea was to see if yum could be used to overcome the distribution barrier in a live environment. And unfortunately - it couldn't. So what to do about it.... Well - ask me anything about parallel processing, mobo development, open firmware, networking or kernel issues and I can most likely answer you. Python? Not my language yet. If yum had been coded in C I would say - link it statically to make it self contained. Very likely we would then have other problems. Python like so many other languages solve these problems by being cross platform, or so they say. Like asm, C and any other language it is cross platform ..... given enough support. You found yourself that possibly the xml passing could be done statically ( not using libxml ). I thought of that. The problem with libxml is that it sucks in this context. It depends on to many other things. LibC for one. Granted - updating from one distribution (more correctly probably: major revision) to the next is the most demanding attempt - because the developer never intended that. Or at least it seems so. I am not picking at YDL here - they follow FC. So I was more looking for: yes or no to that you see the problem and - well - then we will have to think for a while. Maybe an idea will pop up. Karsten
