[arch-projects] [initscripts] crypttab support
seblu at seblu.net
Sat Jul 21 19:07:03 EDT 2012
On Mon, Jul 16, 2012 at 1:25 AM, Tom Gundersen <teg at jklm.no> wrote:
> Furthermore, we deprecate the support for our old crypttab syntax
> (with a warning at boot), but retain the functionality for
> compatibility (in the same way we still have support for the old
> network syntax, this is likely to stay around for a while).
This syntax is frozen since years. A new one provided by systemd its emerging.
Well deprecation seems a good idea but I think you do it too loudly.
Let users choose it's not blame those who use the old one by warning at boot.
Everyone love a bootup without be remembered that you are outdated.
You make something wonderful with initscripts by allowing a smooth
transition between rc.d and systemd.
This allow people (which want it) to have a standardized
configurations for both initscripts and systemd.
If the rise of systemd continues, it's not cryptsetup will be
deprecated, it's initscripts. So my advice is to do it quietly.
> While the syntax is different, the functionality does not change much.
> We gain a few minor features and lose the ability to read a decryption
> key from a removable device (i.e. one that needs to be mounted by the
> cryptsetup code). People who want to use this feature are advised to
> stay with the old syntax for now.
One more reason to not print a warning at boot ;)
> The reason for making this change is mainly that we want to share the
> syntax of crypttab with as many distros as possible, in the same way
> as the syntax of fstab is shared. Furthermore, as systemd becomes more
> and more popular amongst arch users and devs it is a benefit that we
> don't need to keep around two syntaxes for the same functionality.
Just froze the old syntax. Add new feature to the new one (Leannert do
it for us) and people will move smoothly to systemd.
You already plan to support the old syntax, so to don't remove the
code, why tell to people they are on the old syntax.
> Finally, I was never very comfortable with maintaining this code, as I
> hardly use it and it is rather fragile. With this move we are able to
> outsource almost all of the cryptsetup handling to systemd and should
> benefit from their testing and bugfixes.
Just froze it and mark it as no more supported in manpages / bugs report.
> When this eventually moves to [core], I'll put a news item
> highlighting the change to avoid to much worry about the deprecation
This thats "status -v" issue broke the boot of testing users in your
Sébastien "Seblu" Luttringer
More information about the arch-projects