Why jigdo template files for Debian CDs are bigger than necessary
Hi,
this is mainly for Manty, who seemed to be interested in such things. :)
But maybe others will also find the info useful.
While creating some Debian jigdo files, I noticed why ATM jigdo-file 0.6.9
still fails to find many files: Lots of (non-i386) floppy disc images (e.g.
rescue.bin) start with 1k of zero bytes. This causes jigdo-file's internal
"prospective file matches" queue to overflow, so the files are unlikely to
be found.
Maybe you'll remember that I reduced the default --min-length value to 1k
for jigdo 0.6.9, in order to catch files which were shorter than the
previous 4k limit. Unfortunately, this now prevents the floppy images from
being found, grr.
This also explains why there are such differences in the sizes of template
files between different arches; the floppy images of some arches have data
in the first 1k - these images will be found, so the template file of the
CD containing them will be shorter.
I think I might be able to /partially/ solve this issue by changing again
the heuristics used by jigdo-file to decide which matches to drop from the
queue, in such a way that matches starting at 2048-byte boundaries are
favoured.
But the only real way of solving this is to ask the boot-floppies people to
write a few random bytes in the first 1k of each floppy image. I'll ask
whether that's possible.
Cheers,
Richard
PS: I've found a bug in jigdo-file which leads to corrupt template files.
0.6.10 or 0.7.0 will hopefully be out soon.
--
__ _
|_) /| Richard Atterer
| \/¯| http://atterer.net
¯ '` ¯
Reply to: