On Wednesday 03 November 2010 13:33 Petter Reinholdtsen wrote: > [Holger Levsen] Hi, > > we could solve the DVD space issues by simply dropping the idea of > > having multiarch DVDs and instead publish 3 DVDs, one i386, one > > amd64, plus source DVDs. > > > > What do you think? > > I think we should evaluate the cost and benefits of doing this, and > make a decision based on that knowledge. What problems and > advantages for the users, administrators, distributors and developers > do you see for the current approach and a splitted approach. Perhaps > something for a wiki page to gather the arguments for and against? I would like to have that informations gathered on a wiki page. > Why is it that no-one is willing to spend time evaluating the > packages we install today, to see if they all make sense? One point here is for sure, that people do only know that about packages, they use for their own No one alone is in my opinion able to say for every package, if it is useless or not. This is what every user should decide for his own purpose of use > We have duplicate functionally all over the place, but no-one seem > willing to spend the time to figure out which package are good and > which are not good enough to be part of the default installation set. > We get random single proposals like dropping blender, but no > structured evaluation of the packages we install. How many chemical molecule visualizer do we need? Which ones should we keep? How many > mathematical graph drawer applications do we need? How many video > players do we need. > > All these are questions raised by me on > <URL: http://wiki.debian.org/DebianEdu/Status/Squeeze >, but no-one > seem to be interested enough and have free time enough to work on it. > Perhaps we should just set a cutoff point based on popcon.debian.org > numbers? Perhaps we should only keep packages that are translated to > our core languages. Perhaps we should drop those pulling in old a > lot of libraries. Perhaps we should combine these and other criteria > to generate a score for each package and use that score to drop > packages. I do not know and do not have the spare time to spend on > it. But I do know we install some crappy software and heaps of > duplicated functionallity, and believe we should start there to > increase the quality of the Debian Edu distribution. Only the summary of all users minds, qualified by their purposes of use for the package of interest in addition of other factors like the amount of users are using it, or the availability of translation to our core languages and so on, will give a basement for the decision to put on the dvd or not. But without getting users, telling us their minds, we are in fact unable to get this basement and need to decide on facts we have and our own minds. On the other side, how much work, cost, or what else is needed, is it, to provide architecture depending DVDs? I see the best solution in beginning to provide architecture depending DVDs (if it doesnt fails on the point above) _and_ starting to get the list of usefull packages consolidated based on the popcon.debian.org numbers. And afterwards begin with getting more criterias into that decision process. Greetings, Jürgen Leibner -- juergen@leibner.eu GnuPG Key ID: 0x9CD43E5A Finger-Print: D0BC A628 A265 2FF9 A9CC 4D69 F59C 59BA 9CD4 3E5A
Attachment:
signature.asc
Description: This is a digitally signed message part.