Can we get rid of the ext fs?
Since I think the ext2 fs predates SLS, I think it is a good bet that
anyone who is still using ext must be a pretty good Linux hacker and
probably reads this list. Thus I think it is safe to assume that if
no one says "I am using it" in response to this message, no one is.
FWIW here are some of my rambling thoughts on removing unneeded code from
the kernel.
1) For the even numbered kernels, if the configure script presents
the user with an option, it should be known that that option
works. For example, does anyone know if XT harddisk support,
or the ext fs even work in the 1.3.xx kernels?
2) Things like XT harddisk support, the xia file system, and the
old SVR3 fs, which should work, but have never been tested on
non-intel platforms should not show up in the configuration
for these platforms (at least for the even numbered kernels).
3) Some generally accepted policy for removing subsystems from the
kernel needs to be agreed upon. Here is one way it could be
done:
A) When a new even numbered release is about to be made, a list
of potentially unnecessary subsystem is created. This list
should be quite liberal.
B) The configure script for the even numbered release is altered
so if a users elects to configure a subsystem on the list,
they are presented with a message which tells them that that
particular feature is going to be discontinued in the next
release unless Linus is informed that someone still needs it.
They will also be given information on how to get a message to
Linus via email, snail mail, or telephone through some
organization such as Linux International (I dont figure Linus
wants everyone to know his phone number and end up getting calls
at 10PM EST).
C) In the odd numbered kernel immediately following the even numbered
release, the subsystems on the list are removed. As messages
come in saying "I'm using an XT controller on my AlphaStation",
these subsystems can be restored.
Thanks,
Jim