...is plenty stable and should be merged into the 2.4.24 kernel.
15% (301 votes)
...should have been merged into 2.4 a long time ago.
21% (423 votes)
...is too intrusive, and shouldn't be merged into the mainline 2.4 kernel.
6% (115 votes)
...is a great 2.6 feature.
17% (335 votes)
...should only be merged if Marcelo and the various maintainers think it should go in.
27% (535 votes)
...are three letters that together don't mean anything to me.
15% (309 votes)
Total votes: 2018
SCO issue arrises
Well like some people pointed out. XFS and JFS are two of the things that could probably UNIX IP infringments. They probabaly aren't, but until that is clear it should not be included.
System V
System V doesn't include any journaling filesystem. Nor does it include ACL support or other advanced features. The only way you could consider them tainted is that there were originally implementations for AIX and IRIX. So what. IBM and SGI wrote the code, they can GPL it if they want.
SCO can piss off. And if you're trolling, you can too. Otherwise, try to inform yourself before you make comments.
BTW, this doesn't stop Linus from including these in 2.6. And Marcello is well out of SCO's grasp, being a Brazilian.
-molo
Migration prompt
I think XFS shouldn't be merged. It's a great prompt for migration to 2.6.
This would help making 2.6 really stable, because more people would test 2.6.
--
:wq
Thats it !
yes , many people want to merge XFS in 2.4 , they have right to request but XFS is 2.6 feature ...
i think after that 2.6 is *REALLY* Stable , its right to merge XFS into 2.4 , maybe in the middle of 2004 ?
Risks vs. possible gains
Why is it such a problem for the people that want/need to use XFS for whatever reason to patch it themselves? Why does it need to be part of the "mainline" kernel? In my opinion the risks far out weigh the benefits. When you look at the total number of people using 2.4 compared with the number of people interested in getting XFS merged the risks just don't make sense.
But removing a 'feature' like
But removing a 'feature' like the OOM killer isn't risky?
It's a problem to patch in XFS ourselves because it makes installs a PITA because virtually no install CDs have the patches applied. so you either make your own custom install CD (or floppies) or install to ext3, install an XFS kernel, make a new XFS partition and copy your installed files there.
And what risks are you talking about? I've been using XFS as my rootfs and on a handfull of other semi-large filesystems since the Linux port hit 1.0 (2001 IIRC) and I havn't had a single oops or hang that I could blame on XFS and so have tons of other people. I'm not saying there's 100% no chance XFS will cause a problem for someone, but every patch has the chance of causing some unintended problems.
XFS in installers
virtually no install CDs have the patches applied
Wrong. Slackware could be installed using XFS.
--
:wq
I said virtually, meaning alm
I said virtually, meaning almost, not quite or nearly.
And quite frankly I wouldn't use slackware if you paid me to, no matter how many kernel patches they apply.
...lot's of distro's
Well, take Knoppix then.
Uhhh..
3rd party Debian XFS CD exists. Gentoo can do it too. RedHat has done it for quite a while.
RedHat has never done it, SGI
RedHat has never done it, SGI releases patched RedHat CDs but they're not supported by RedHat.
Comparing risks
Well, of course, every major change bears some risks. But then, you have to say, XFS is one of the best-tested file systems available for Linux, isn't it? The patch has been around for years, now, most distributions nowadays have it included, anyway, and XFS is widely deployed.
Another point is that folks do seem to be less conservative when the thing in question is not called "XFS". E.g. why has ext3 been included in 2.4 kernels? Compared to XFS it is much less mature, much less tested and it actually did fuck up more than one Linux server I support (you know this "I/O error: Invalid argument" problem, I guess, you know ext3 randomly corrupting data, as well - problems I never had with XFS or even with plain old ext2). So what risks are you talking about?
Of course, we all can patch our kernels. But this actually applies to every single feature. Why having Appletalk/Bluetooth/IRDA support in the mainline kernel since the total number of people using these features compared to the total number of people using 2.4 kernels is low. Let them patch their kernels if they do need this stuff.
xfs will be in 2.4.24
Marcelo's post to lkml: http://lkml.org/lkml/2003/12/8/35
Re: I Think That XFS
is a great font server. Keep up the great work Keith! Good luck on Xouvert...
lol
lol