Inclusion of ext2fs patch in Debian's hurd package
Hi,
I've read a discussion in #hurd channel about the inclusion of ext2fs
patch in Debian's hurd package. Below I write about my opinion about this.
First of all, I don't recommend making patched ext2fs to be the default
until there are problems with it. Currently, Jeff Bailey reported a
problem that I still haven't tried to reproduce, but I think I'll
reproduce it. This month is very busy for me and I can't give promises
for timing.
*But* I recommend it to be included as an alternative to current
ext2fs.static. Perhaps ext2fs.static.big can be registered as a lower
priority alternative to ext2fs.static. When there are no known problems
with the patch, we can make it the default alternative.
So I think it's good if next hurd package upload includes
ext2fs.static.big as an option to those that want to test it.
Concerning acceptance in upstream Hurd: AFAIK Roland and Marcus haven't
read the patch and so they didn't give opinions. Thomas (and Roland?)
has its own ideas how it should be done, but I don't remember if he told
sth against my approach (I could be wrong -- should be checked in mail
archives). My approach is implementing block cache layer used to access
metadata blocks. This is the natural approach in UNIX variants but, of
course, there it is implemented in the kernel, not in user space. Neal
has given some valuable technical advices, but he hasn't read the whole
patch, so no strong opinion (positive or negative), again.
My plan about promoting the patch to mainline is simple:
1. There should be no known problems with the patch.
2. Mail RC2 to bug-hurd with inlined documentation (like the one in sv),
changelog and the patch itself. This is to ease Thomas and Roland.
Regards,
ogi
Reply to: