On 08/23/2010 01:04 PM, Chris Lumens wrote: > Initial testing of the giant initrd is not going so well. First, it > appears you have to bump the memory up to 1 GB. Second, qemu and the > kernel are both pretty unhappy about such a large initrd. Which compression format is used where? Output from lzma often is 15% smaller than "gzip -9". > -r--r--r-- 1 root root 99158752 Aug 23 14:05 usr/lib/locale/locale-archive Several possible cuts: Take the top 10 locales by smolt listings. Or, take the six official languages of the United Nations (English, French, Spanish, Russian, Chinese(Mandarin), Arabic). Or, set a limit of 30MB and take what fits. Or, take only _your_ favorite locale. > -r-xr-xr-x 1 root root 2280024 Jul 20 01:29 usr/lib64/dri/radeong_dri.so [snip] > -r-xr-xr-x 1 root root 2244696 Jul 20 01:29 usr/lib64/dri/r600_dri.so > > Are these really necessary for a responsive installation experience? > That's 30 MB right there. xdriver=vesa nomodeset > -r--r--r-- 1 root root 8618994 Jul 21 21:03 usr/share/cracklib/pw_dict.pwd > > We only use this 8.6 MB file to check root's password once. There's got > to be a smarter and smaller way of doing that. If required to be self-contained (download not allowed) then you can change the algorithm to a hashing scheme that identifies "possibly weak" [subject to both false positives and false negatives], with a warning to check later. -- _______________________________________________ Anaconda-devel-list mailing list Anaconda-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/anaconda-devel-list