> I see there a line with "Content-Type: text/html". Sometimes it is on top > and sometimes on bottom. I used the IE6 to view the page. It's maybe a wrong > http-header or something similar. > I'm afraid you're going to get that with every search and to be honest, I'm not sure why it is. The Content-Type line is always the first thing that the app spits out. If the browser hasn't already figured out what type of content to use, it'll have issues if I don't send this. If, on the other hand, it's already interpreting tracker.py's output correctly then it just prints the Content-Type string verbatim. In other words, if I leave it out the app breaks, but if I leave it in, sometimes the text you see gets printed. If anybody has suggestions for how I can have Tracker differentiate between the two cases, I'm all ears. > Maybe some support could be added in tools that are involved in installing > the packages. My fantasy: You can directly search for packages in the > repositories and download and install it in a go. Perhaps some other > developer-teams are willing to do this in their projects. I've been toying with these ideas as well. There is an overhaul of system-config-packages in the works and when I have some time to do so I'd like to contribute to it. Likewise I was playing with ways to tie yum into the Tracker. This would be simple in one sense because Tracker can generate yum.conf files and yum allows you to point it to a url for its config. But it would also be problematic since yum wants to run as root to install packages and the first time someone did a yum install from the config file provided by tracker it would probably take a *very* long time for headers to download. Anyway, I haven't talked to the people involved with either project yet, but both have been on my mind. Like I said, perhaps when I have some more time to devote to it. And speaking of my Real Job... --Brad
