better mksquash released at sources

[nick65go]

from http://sourceforge.net/projects/squashfs/files/squashfs/squashfs4.3/
squashfs4.3.tar.gz    2014-05-12    182.6 kB

"Welcome to Squashfs 4.3. This is the first release in over 3 years, and there are substantial improvements to stability, new compression options and compressors, speed optimisations, and new options for Mksquashfs/Unsquashfs."

I used just for testing the NEW option "unsquash -stat xorg-server.tcz"

"Found a valid SQUASHFS 4:0 superblock on /mnt/sda3/tce/optional/xorg-server.tcz.
Creation or last append time Sat Nov 30 12:35:54 2013
Filesystem size 3283.15 Kbytes (3.21 Mbytes)
Compression gzip
Block size 131072     <-- so the recommended 4096 KB block is not used all the time 
Filesystem is exportable via NFS
Inodes are compressed
Data is compressed
Fragments are compressed
Always-use-fragments option is not specified
Xattrs are compressed
Duplicates are removed
Number of fragments 8
Number of inodes 55
Number of ids 3"

I cheated by unpaking from archlinux site into corepure64 the squashfs-tools 4.3-3 and its dependencies drooped into /lib (lz4,  lzo, xz,  zlib)
the new vesrion of squashfs-tools works with the same squash 4.0 format used by kernel, just the tools are improved.

The idea is that for corepure64 we are not limited by CPU speed (otherwise why you choose 64 bit version?). So the squashfs format could use better compression, like bigger blocks and gzip compression-level should be  by default 9.
If you are adventurous, rebuild the kernel module with lzma support and use xz (the best compression!) in tcz extensions such as the big libreoffice.tcz

Anyway, it is useful to see the used block of already created tczs in the repository, by using ./unsquash -stat my.tcz

[curaga]

The block size influences memory usage more than cpu. You're right though that 64-bit systems would likely have a lot of RAM too.

[nick65go]

For core 32 recomended block size is 4kb. What about recomended squash bock size for core64, 128kb?

New version of squash tools could be compiled without additional compresion suppoort options (lz4,  lzo, xz) but just as they are now with zlib only. Could we take advantage of new stability of the new tools, so to do not pach anymore the squashfs kernel module in the core base, so we get near a vanila kernel surces?

[curaga]

The only patch to the squashfs module we currently have is purely cosmetic?

For the blocksize, I see no reason not to use the highest, 1mb. But this does need more discussion before you should submit such extensions.