I miss slax modules' outstanding compression ratio given by squashfs-lzma, and for some large personal tczs I don't care about reading speed.
Not being smart enough to patch a kernel to support squashfs-lzma, I experimented a bit with fusecompress, and found that combining "mksquashfs -no{I,D,F,X}" with "fusecompress -o fc_c:lzma,fc_b:256" gives compression ratios just below that of slax's lzm.
So I remastered microcore.gz, just extending tce-load line:
[ "$?" == 1 ] || sudo busybox mount "$THISAPP" /tmp/tcloop/"$APPNAME" -t squashfs -o loop,ro,bs=4096 2>&1
into
[ "$?" == 1 ] || sudo busybox mount "$THISAPP" /tmp/tcloop/"$APPNAME" -t squashfs -o loop,ro,bs=4096 2>&1 || \
{ { sudo mkdir /mnt/fu; sudo fusecompress "${FROMWHERE}" /mnt/fu ; true ; } && sudo busybox mount "/mnt/fu/${APPNAME}.tcz" "/tmp/tcloop/${APPNAME}" ; }
which basically turns the directory containing a fusecompressed tcz into a fusecompress host directory mounted in /mnt/fu, then loop-mounts the virtually uncompressed squashfs tcz resulting in /mnt/fu, and proceeds as for a standard tcz.
Note that you have to specify "-o fc_c:lzma,fc_b:256" just upon creation of extension, not when reading it.
I am currently testing it, seems to work.
Now my questions/doubts: