Am Di, 28. Sep 2004, 12:05:31 -0400 schrieb Joey Hess: > As bast I can tell, the patch in this bug does not change the behavior > of rootskel. I cannot apply patches without an explanation of the bug > that the patch fixes, and there is no explaination of why the changes in > this patch are useful. The problem is that setting debian-installer/framebuffer=false on the kernel command line does not have any effect, at least on sparc. Also, debian-installer/framebuffer is somewhat misnamed, as it does not control whether a framebuffer is used at all, rather it controls whether to use a multilingual terminal (such as bterm) on top of the framebuffer. That is how I read the description of debian-installer/framebuffer in the Installation Manual, section 5.2.1. "Debian Installer Parameters". > The patch makes TERM_FRAMEBUFFER not be set if it is already set. But > the only thing in all of rootskel that sets TERM_FRAMEBUFFER is the code > here, so this is a no-op. The user could set TERM_FRAMEBUFFER=false on the kernel command line to circumvent the autodetection code. Of course this is not documented anyway, and invites confusion like DEBIAN_FRONTEND vs. debconf/frontend in the future, so we should probably drop this test. > The patch adds a check for debian-installer/framebuffer=true and does > not set TERM_FRAMEBUFFER if it's not. But this duplicates the check made > in S40framebuffer-module-linux-*. If debian-installer/framebuffer=false, > then S40framebuffer-module-linux-* will not load framebuffer modules, > and so /dev/fb/0 will not exist, and so the existing check in > S35framebuffer-linux should suffice. S40framebuffer-module-linux-* does only exist for alpha, mips, and x86. At least sparc64 has the framebuffer compiled in (I suppose there is no text console on sparc). But the sparc framebuffer does not work with bterm. Setting debian-installer/framebuffer=false is the documented workaround, but it also does not work on sparc. Have a look at #272552 and #272562 for a detailed description of the failure. > Posting a proper explanation would have been a much better way to get > this fixed than posting an unexplained patch. At this point all I know > is that debian-installer/framebuffer=false works on i386, and may not > elsewhere, but I can't tell why not. It could even still be user error; > this string is known to be hard to type correctly on eg, French > keyboards. I hope this explains it enough, please ask again otherwise. Jö. -- Q: How does a Unix guru have sex? A: unzip;strip;touch;finger;mount;fsck;more;yes;umount;sleep -- unknown source
Attachment:
signature.asc
Description: Digital signature