0.14.9

Minor bug fix (installed zsh scripts); and README clean up.

Diff/stat

v0.14.8

Minor bug fix (location of usrdir of installed building scripts.)

diff/stat

v0.14.7

Minor bug fix (suspend/resume) and improments: added hooks/mkswap-zfs and scripts/suspend. The latter is little helper to soften the pain of using swsusp/TuxOnIce implementation.
The short log and diff are over there.

v0.14.6

Minor bug fix (diff.)

v0.14.5

This a bug fix (bcache) and mdev clean up.

• (Trivial) fix of bcache hooks and added basic support in init
  (ibcache argument will properly execute the hook);
• Cleaned up mdev scripts mess and added dm_link
  to take care of device-mapper link (in /dev/mapper);
  apparently, lvm has a hard dependency on udev
  (it does not matter with 0.14.4 updates and this one);
• Reverted putting shm on ext4 zram backed device.

The complete diff is over there.

v0.14.4

This a bug fix only.

There mainlay three annoying fixes:

• Fix broken -k|--kv, -d|--usr, -c|--compressor after a short lived fix of -k|--kv,-u|--usr;
• Fix udev hard dependency to lvm (why should a user be forced to use udev?!);
• Fix stting properly configuration file for installed variant.

The complele log and commit list is over there.

v0.14.3

Bug fix release:

• Fix the annoying prefix slash '/' in squash dir related script (the previous measure in 0.14.2 was not enough);
• Fix setting mode (if present in zram init script).

Diff stat is over there.

v0.14.2

This a feature release only... minus a squash directory late breakage (squashdir-mount init script) which preventing it to unmount aufs and squashfs file systems. An easy fix. Wait, I've just had enough of it and went to fix/update possible slash (/) or double slash issue. (This due to the fact that squashed dirs could be passed without the leading slash while the root directory should had the the leading slash. Enough with that, now a leading or missisng slash is supported watsover the variable.)

The most interesting stuff is that, the building script support an env(ironment) array where every kernel cmdline argument and more can be defined. As a result, a default profile ca be built. And init script will pick up if present. (And then one can boot with an empty cmdline; and forget fighting against GRUB or other boot loaders with a lengthy cmdline.)

There is more of it. ienv kernel cmdline can be used disable all of it or part of it: e.g. ienv=:ikroot:iroot will disable both variables and leave the other.

ZFS hook was re-written and refactorized, so it will be run if and only if izfs cmdline argument is present.
Improved dmopen() (raid) function to be able to parse multiple cmdline variations (short to long).

Chekout the ChangeLog/README and configuration file for more info.

All of this comes at the cost of only 24 (+230/-206) lines with the help of a little clean up (zfs+README) which is impressive taking into account the amount of lines involved.
More details in the diff stat.

v0.14.1 (NO FALSE PROMISE)

This was supposed to be a features release only (with a few improvements) and annoying bugs get in the way.

Notable fix is imod kernel cmdline argument which was useless since...;
Another important one is not to double imopt valude (the old parsing way did that);
And fix a colision when ro|rw and imopt are present in kernel cmdline;
And a bunch of other minor fixes (see the commits).

New features:

Added kernel module group support in boot group, so now boot group has kms group by default. And this is a sane and wanted default to not have to keep imod=kms forever.

There is a default font and keymap to not have to keep a ikmap=keymap:font forever; And the default is the first font/keymap found. So append font/keymap wisely with -f|--font or -y|--keymap, or else, define wisely opts[-font] and or opts[-kmap] or even opts[-f] or opts[-y] to be sure to be first.

Last notable feature is the ability to copy kernel module dependecies on the fly... This is an old wanted feature. This feature ended up by getting rid of find when looking for modules which reduces disks seeks.

v0.14.0 aka ALL FUNCTIONAL

Well, this was supposed to be a minor (v0.13.8) bug fixes release, it turned out to be quite big (nothing scarry--there are way too many single line or two commits).

Only init service scripts get it right and should be solid as rock. bcache script get a quick clean up/update to get something usable (still need two cmdline argument to be used: ibcache=somthing ihook=bcache).

The real change is in init/functions/README.texttile slight clean up and update with a slight improved api like for idebug, imodule transformation into ihook, arg() usage limitation in favor of set-ting positional parameters, zfs related hooks, zram hook re-write following almost a zram.initd re-write. Well, quite a big change, hence the long commit list.

Everything should (really) just works! (That's what I've said in the previous release note and whenever new changes get in... some breakage appears. The real annoying breakage was in CONFIDIR broken string introduced the 0.13.x series which could lead to unbootable system.)