On Thu, 15 Jun 2000, Matthew Hawkins wrote:
> On 2000-06-13 16:32:02 +0100, James Sutherland wrote:
> > On Tue, 13 Jun 2000, Hans Reiser wrote:
> > > If you can reproduce a bug on the current version of reiserfs, please
> > > do so and report it. I think you'd rather throw mud though.
> >
> > No, I'd rather just leave RFS to mature for a while, while we work on
> > getting 2.4.0 ready.
>
> Hmm.. so what you're saying is you don't have the time or ability to
> reproduce bugs in reiserfs and maybe help solve them, but you do have
> time to respond to every post on the inclusion-of-reiserfs thread saying
> how buggy it is?
I am not "saying how buggy it is", I am just saying it shouldn't go into
2.4.0.
> > I've heard a couple of horror stories about RFS chewing up data.
>
> I've heard that if I send $5 to three addresses in lamerica, and add my
> address to the bottom and send the letter I got from these people to
> everyone in my address book, I'll become a billionaire!
>
> "It must be true, I heard it on the internet!"
If you heard it directly from someone you knew and trusted, you probably
would give the report some weight.
> If you have not experienced, and do not have the time/ability to produce
> bugs in reiserfs, then it doesn't take too many brain cells to figure
> out that in your own experience, there are no bugs in reiserfs.
>
> So stop spreading your FUD to this list.
This would be the "FUD" that Chris believes and is currently working on?
> Feel free to reproduce a bug (not only in reiserfs, in anything!) and
> complain to the list. Include a list of steps that will reproduce it.
> Hell, include a patch that fixes it.
>
> If you're *really* concerned about getting 2.4.0 ready, that is what
> you'll do. From what I've seen (and, I must say, I read this list from
> archives usually 2 days old) all you seem to do is spread FUD about
> things you don't like.
You've obviously missed quite a bit, then. Possibly because the thread in
which Chris agreed with me about ReiserFS not belonging in 2.4.2?
> I don't even use reiserfs, but (like Voltaire) I'll defend to the death
> the right of any driver at an acceptable level of development to be
> included.
Not straight into a stable kernel series during a code freeze, I hope.
> I'm one of the many here who have said Y to the experimental new
> second extended filesystem not all that long ago, which hasn't gone
> without its own fair share of post-inclusion development and bug fixes
> in the years since.
Of course ext2 isn't perfect - no code is. However, would you have
supported the idea of nailing it on the end of 2.0.0, when it hadn't shown
up in any of the previous development releases?
> > They'd probably be less happy if they'd just lost an entire SQL database
> > to your software...
>
> I've lost three years worth of archived email to a filesystem corruption
> bug in ext2. I've lost the latest development version of software I
> write to a corrupted ext2fs.
>
> Back. Up. Your. Data.
>
> (like Linus, I prefer to simply put the important stuff up on ftp and
> let the world mirror it ;)
I'm sure we've all lost data to occasional random problems - but this
looks, so far, like a rather more serious thing. Postgresql uses a
perfectly standard API to access files, as a result of which ReiserFS
apparently hoses data. Chris is still looking into this one, so I don't
know all the details yet - but it is certainly serious enough to warrant
investigation.
> > The integration into the kernel doesn't appear to work properly yet, FWIH.
>
> Hmm.. again from what you've heard. Maybe what you've heard is as out
> as date as my own previous comments "from what I heard" on the LSB, and,
> like that, is actually in a far better state than you last heard.
One of the people who agrees with me on this is Chris. Presumably you will
now claim he is out of date with his own code?
> > Even if it does, it's too late for the code to go in now, so what are you
> > arguing about?
>
> So was S/390 support, so are these VM tweaks. Look through changelogs,
> a lot of stuff goes in "late". Look in past history of kernel
> development, a lot of stuff goes in late. If the world doesn't run to
> Linus Torvalds' timetable, it certainly doesn't run to yours.
The S/390 inclusion doesn't affect existing platforms; the VM tweaks are
needed bug-fixes to an existing system. ReiserFS is neither.
> You know what to do if you don't want reiserfs? When you configure your
> kernel and the option for reiserfs is presented, say no. Don't be
> scared, lots of other people will be saying "no" too. It's not
> difficult. It's even easier than saying "yes" because it'll no doubt be
> the default.
>
> reiserfs contains no nicotine. You're not under any mind-altering
> chemical making it biologically difficult to say no.
>
> Just like other kernel options you don't want, you can say no.
>
> Repeat after me:
> Include support for EXPERIMENTAL reiserfs? [y/N] n
>
> Denying the right of people who may want to say "y" to do so will lead
> to the Dark Side.
Try convincing Alan and Chris of that.
James.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Jun 15 2000 - 21:00:35 EST