Hi Andreas, Andreas Tille <[email protected]> writes: >> - the number of defined "teaser" packages may be higher than the >> available space on the web page, so it would need to select the >> "highest ranking" (or a random selection?) > > We could cut at a defined maxteaser value. The only problem I see in > this approach that the manual maintenance becomes a burden if you need > to ensure manually that no duplicated Teaser values occure.
This seems to be too error-causing to me. Then (if we adopt the "teaser" tag) it would probably be better to have them unsorted. >> * Do we want to have them consistent between possibly different places >> (Web page, tasksel)? Or is a local definition better here? > > Does tasksel allow for a certain sequence? Not that I knew of. It is more a think-play (what is english for "Gedankenspiel") to see where the teaser definition really belongs to: is it part of the presentation (and should be defined in the web page), or is it an attribute of the package itself? > A totally different option which could live without the Teaser value > would be to define the key packages according to their popcon value. > This could serve as kind of an objective measure to rank packages in a > task and it is implemented right now. I'd be against this since the popcon does not reflect what I thought for the teaser: It is not just "What is mostly used?", but "What could attract people?". In my specific case (debian-astro), the two most popular packages are f.e. kstars and stellarium; I would however select only *one* of them as a teaser. The other would attract almost the same people. I would instead try to choose three packages for three audiences, like: * educational (that's kstars, or stellarium) * amateur (indi???) * professional (most likely astropy, even if does not have the highest popcon) (or take kstars for the first two categories, and take another professional program for another field of astrophysics). The freedom of choice in the presentation is something that should IMO be kept. Best Ole
