Holger Levsen wrote:
Hi, On Thursday 28 September 2006 21:41, Attilio Fiandrotti wrote:I spent last days in collecting reports from Debian PPC users (thanks to all those guys who kindly provided these datas) and now i've reported the results in this wikipage [1].Great.A wider testbase is needed tough to validate existing results: pheraps i should issue a call for testing on debian-powerpc ?Definitly!It would be helpful if you could also add brief instructions how to test (what to download where, how to boot) on that page. In October I will be able to test on a pismo g3 notebook, an imac g5 and hopefully also an old g3 imac, though I'm not sure if I can get RAM for it, currently that machine only has 32 MB.
I did one more test with a friend's MiniMac and it turned out the g-i works if linux_input module is disabled. As a general rule, the linux_input directfb module has to be disabled, otherwise Mac boxes crash. The only exception to this general rule are "PowerBook6,7" iBooks, where linux_input *must not* be disabled. One nasty, unexpected, thing i noticed is that keymap results completly messed up after a keymap change by language-chooser (or whatever d-i component takes care of performing this task). So, it seems major issues on PPCs are related to DFB's input modules, not to graphical drivers (disabling hardware acceleration and forcing off / removing gfxdrivers is ok for all known test cases). I'll prepare a list of questions for PPC testers, then i'll issue a call for testing on debian-powerpc.
Right now g-i on PPC seems to be quite buggy, to be optimistic.. :( Attilio