Reiser4 status as of this weekend

Reiser4 status as of this weekend

[Hans Reiser]

Russell Treleaven wrote:

>Hi Hans,
>
>Your website has been saying "Reiser4 is in final testing, and will ship
>soon!" for a long time.
>Terms like long and soon are kinda vague. How long has the website
>stated that Reiser4 will ship soon?
>I really like the idea and wan't to try out Reiser4.
>
>Sincerely,
>
>Russell Treleaven
>
>
>
>  
>
Some users are saying that it is rock solid (please contradict dear 
readers).  We the developers can use iozone to make it crash by mmapping 
more than all of ram and dirtying all the mmapped pages.  Linux is not 
well structured for managing memory under that circumstance.   The clean 
fix for that for reiser4 given the way the Linux design is,  is 
substantial work to do right, and is in progress.  We currently have a 
fix that costs performance, and are fixing the performance issue (which 
is fixable but requires code....).

We would probably have it ready to ship this week, but Nikita is leaving 
Namesys after having done a great job here for way too little pay, and 
that means we need to have him and Saveliev spend a full week doing a 
code review before he goes, and Zam and I are busy earning the next 
month payroll money by doing work for digeo, and such is the free 
software "business".  Sigh.

So, stay tuned, and check again in two weeks.

Hans

Re: Reiser4 status as of this weekend

[Redeeman]

NOOOOOO NIKITA! DONT LEAVE!!

On Sat, 2004-05-15 at 23:45, Hans Reiser wrote:
> Russell Treleaven wrote:
> 
> >Hi Hans,
> >
> >Your website has been saying "Reiser4 is in final testing, and will ship
> >soon!" for a long time.
> >Terms like long and soon are kinda vague. How long has the website
> >stated that Reiser4 will ship soon?
> >I really like the idea and wan't to try out Reiser4.
> >
> >Sincerely,
> >
> >Russell Treleaven
> >
> >
> >
> >  
> >
> Some users are saying that it is rock solid (please contradict dear 
> readers).  We the developers can use iozone to make it crash by mmapping 
> more than all of ram and dirtying all the mmapped pages.  Linux is not 
> well structured for managing memory under that circumstance.   The clean 
> fix for that for reiser4 given the way the Linux design is,  is 
> substantial work to do right, and is in progress.  We currently have a 
> fix that costs performance, and are fixing the performance issue (which 
> is fixable but requires code....).
> 
> We would probably have it ready to ship this week, but Nikita is leaving 
> Namesys after having done a great job here for way too little pay, and 
> that means we need to have him and Saveliev spend a full week doing a 
> code review before he goes, and Zam and I are busy earning the next 
> month payroll money by doing work for digeo, and such is the free 
> software "business".  Sigh.
> 
> So, stay tuned, and check again in two weeks.
> 
> Hans
-- 
Regards, Redeeman
()  ascii ribbon campaign - against html e-mail 
/\                        - against microsoft attachments

Re: Reiser4 status as of this weekend

[mjt]

On Sat, May 15, 2004 at 02:45:25PM -0700, Hans Reiser wrote:

>Some users are saying that it is rock solid (please contradict dear 
>readers).  We the developers can use iozone to make it crash by mmapping 

I haven't had problems for a long time.

Then again, I haven't used iozone ever.

>well structured for managing memory under that circumstance.   The clean 
>fix for that for reiser4 given the way the Linux design is,  is 
>substantial work to do right, and is in progress.  We currently have a 
>fix that costs performance, and are fixing the performance issue (which 
>is fixable but requires code....).

Would you say this is the biggest issue?
What about NFS?

>We would probably have it ready to ship this week, but Nikita is leaving 
>Namesys after having done a great job here for way too little pay, and 

!!!
This is terrible news... I'm sorry, I'm at a loss of words...

Nikita, hope to see you on IRC still and thanks for all the great
work and the help you've given me and everyone else.

Best of luck.

>that means we need to have him and Saveliev spend a full week doing a 
>code review before he goes, and Zam and I are busy earning the next 
>month payroll money by doing work for digeo, and such is the free 
>software "business".  Sigh.

This may be a tactless question, but what about the grant for implementing
views? It was insufficient?

Have you also considered making it easier for the community to write code?
I'm not talking about write-access handed liberally to bitkeeper, but
maybe some coding guide and stuff could be more prominently available?

>So, stay tuned, and check again in two weeks.

Best of luck to the rest of you too.

-- 
mjt

Re: Reiser4 status as of this weekend

[Christian]

Hans Reiser wrote:
> We would probably have it ready to ship this week, but Nikita is leaving 
> Namesys after having done a great job here for way too little pay, and 

wow, these are sad news....time to say bye & Thanks! for always being 
very helpful and the just-in-time support on this list.

have a nice time,
Christian.
-- 
BOFH excuse #418:

Sysadmins busy fighting SPAM.

Re: Reiser4 status as of this weekend

[David Masover]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

| Hans Reiser wrote:
|
|> We would probably have it ready to ship this week, but Nikita is
|> leaving Namesys after having done a great job here for way too little
|> pay, and
|
|
| wow, these are sad news....time to say bye & Thanks! for always being
| very helpful and the just-in-time support on this list.

Amen.  Being a hapless user, I gave up trying to say something -- you
said it.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
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=Mbpc
-----END PGP SIGNATURE-----

Re: Reiser4 status as of this weekend

[Hans Reiser]

Markus Törnqvist wrote:

>On Sat, May 15, 2004 at 02:45:25PM -0700, Hans Reiser wrote:
>
>  
>
>>Some users are saying that it is rock solid (please contradict dear 
>>readers).  We the developers can use iozone to make it crash by mmapping 
>>    
>>
>
>I haven't had problems for a long time.
>
>Then again, I haven't used iozone ever.
>
>  
>
>>well structured for managing memory under that circumstance.   The clean 
>>fix for that for reiser4 given the way the Linux design is,  is 
>>substantial work to do right, and is in progress.  We currently have a 
>>fix that costs performance, and are fixing the performance issue (which 
>>is fixable but requires code....).
>>    
>>
>
>Would you say this is the biggest issue?
>  
>
It is the only issue that I know of.

>What about NFS?
>  
>
fixed by nikita recently but not released.  See next snapshot.

>  
>
>>We would probably have it ready to ship this week, but Nikita is leaving 
>>Namesys after having done a great job here for way too little pay, and 
>>    
>>
>
>!!!
>This is terrible news...
>
yeah, I agree.

> I'm sorry, I'm at a loss of words...
>
>Nikita, hope to see you on IRC still and thanks for all the great
>work and the help you've given me and everyone else.
>
>Best of luck.
>
>  
>
>>that means we need to have him and Saveliev spend a full week doing a 
>>code review before he goes, and Zam and I are busy earning the next 
>>month payroll money by doing work for digeo, and such is the free 
>>software "business".  Sigh.
>>    
>>
>
>This may be a tactless question, but what about the grant for implementing
>views? It was insufficient?
>  
>
It supports me and one programmer for the first 6 months.  I did not bid 
my being paid a high enough salary to pay the rest of the company on it, 
so in order to get get beyond phase 1, so I must work 2 jobs while 
having a long commute to one of them and running Namesys.  I have to do 
at least half the work in Phase 1 because we need to start the project 
by doing the architectural work....  Also, it requires that 2/3 of the 
work (by cost) be done in the US, and the moscow embassy has gone into 
total asshole mode and isn't giving work visas to most russians.  Still, 
we will try to get work visas in 6 months.  I hope that it being for DoD 
work will increase the chance of getting visas.

>Have you also considered making it easier for the community to write code?
>I'm not talking about write-access handed liberally to bitkeeper, but
>maybe some coding guide and stuff could be more prominently available?
>  
>
It is in the source code..... we have the best commented source code in 
the kernel (Andrew Morton is the only other kernel person who comments 
code well that I know of).

>  
>
>>So, stay tuned, and check again in two weeks.
>>    
>>
>
>Best of luck to the rest of you too.
>
>  
>
thanks

Re: Reiser4 status as of this weekend

[mjt]

On Sun, May 16, 2004 at 10:30:59AM -0700, Hans Reiser wrote:
>Markus Törnqvist wrote:

[mmap]
>>>substantial work to do right, and is in progress.  We currently have a 
>>>fix that costs performance, and are fixing the performance issue (which 
>>>is fixable but requires code....).
>>Would you say this is the biggest issue?
>It is the only issue that I know of.

How about Copy-on-capture? Is it usable? Does it provide better
performance?

>>What about NFS?
>fixed by nikita recently but not released.  See next snapshot.

I'm trying BK pulls. Yesterday it didn't compile, xfs seemed broken.
Retrying now, after bk -r get.

I wrote a lame script, haven't used it much myself, but it should do
something. Like get a BK pull.
http://mjt.nysv.org/reiser/getreiser.sh
http://mjt.nysv.org/reiser/getreiser.sh.txt

>by doing the architectural work....  Also, it requires that 2/3 of the 
>work (by cost) be done in the US, and the moscow embassy has gone into 

Have you tried other governments?

>total asshole mode and isn't giving work visas to most russians.  Still, 
>we will try to get work visas in 6 months.  I hope that it being for DoD 
>work will increase the chance of getting visas.

That must not fail.
The DoD can't say "please give these guys work visas" just to cover
their own interests? Or this has no effect?

>It is in the source code..... we have the best commented source code in 
>the kernel (Andrew Morton is the only other kernel person who comments 
>code well that I know of).

There was also a tangent on the topic of creating daily snapshots.

This might make it easier for people to come to terms with Reiser4.
People have many reasons to dislike BK, so a daily snapshot patch
against some vanilla system may be a good thing.

However, having broken code in the BK for days without having time to
fix it is not good.

So this would require changelogs and some testing on snapshots so that
every once in a while one can be labeled "stable as hell"

-- 
mjt

Re: Reiser4 status as of this weekend

[Vladimir Saveliev]

Hello

On Wed, 2004-05-19 at 23:50, Markus TЖrnqvist wrote:
> On Sun, May 16, 2004 at 10:30:59AM -0700, Hans Reiser wrote:
> >Markus Törnqvist wrote:
> 
> [mmap]
> >>>substantial work to do right, and is in progress.  We currently have a 
> >>>fix that costs performance, and are fixing the performance issue (which 
> >>>is fixable but requires code....).
> >>Would you say this is the biggest issue?
> >It is the only issue that I know of.
> 
> How about Copy-on-capture? Is it usable? Does it provide better
> performance?
> 

When we moved to mm kernel it broke. When it was working - it did not
show significant effect in quick tests like bonnie.


> >>What about NFS?
> >fixed by nikita recently but not released.  See next snapshot.
> 
> I'm trying BK pulls. Yesterday it didn't compile, xfs seemed broken.
> Retrying now, after bk -r get.
> 
> I wrote a lame script, haven't used it much myself, but it should do
> something. Like get a BK pull.
> http://mjt.nysv.org/reiser/getreiser.sh
> http://mjt.nysv.org/reiser/getreiser.sh.txt
> 
> >by doing the architectural work....  Also, it requires that 2/3 of the 
> >work (by cost) be done in the US, and the moscow embassy has gone into 
> 
> Have you tried other governments?
> 
> >total asshole mode and isn't giving work visas to most russians.  Still, 
> >we will try to get work visas in 6 months.  I hope that it being for DoD 
> >work will increase the chance of getting visas.
> 
> That must not fail.
> The DoD can't say "please give these guys work visas" just to cover
> their own interests? Or this has no effect?
> 
> >It is in the source code..... we have the best commented source code in 
> >the kernel (Andrew Morton is the only other kernel person who comments 
> >code well that I know of).
> 
> There was also a tangent on the topic of creating daily snapshots.
> 
> This might make it easier for people to come to terms with Reiser4.
> People have many reasons to dislike BK, so a daily snapshot patch
> against some vanilla system may be a good thing.
> 
> However, having broken code in the BK for days without having time to
> fix it is not good.
> 
> So this would require changelogs and some testing on snapshots so that
> every once in a while one can be labeled "stable as hell"

Re: Reiser4 status as of this weekend

[Matthias Andree]

mjt@nysv.org (Markus Törnqvist) writes:

> I'm trying BK pulls. Yesterday it didn't compile, xfs seemed broken.
> Retrying now, after bk -r get.

Try "bk -Ur get -S". That's a bit faster and doesn't check out BK stuff,
only the actual user data (read: source).

-- 
Matthias Andree

Encrypted mail welcome: my GnuPG key ID is 0x052E7D95

Re: Reiser4 status as of this weekend

[Hans Reiser]

On Thursday 20 May 2004 02:39, Vladimir Saveliev wrote:
> Hello
>
> On Wed, 2004-05-19 at 23:50, Markus TЖrnqvist wrote:
> > On Sun, May 16, 2004 at 10:30:59AM -0700, Hans Reiser wrote:
> > >Markus Törnqvist wrote:
> >
> > [mmap]
> >
> > >>>substantial work to do right, and is in progress.  We currently have a
> > >>>fix that costs performance, and are fixing the performance issue
> > >>> (which is fixable but requires code....).
> > >>
> > >>Would you say this is the biggest issue?
> > >
> > >It is the only issue that I know of.
> >
> > How about Copy-on-capture? Is it usable? Does it provide better
> > performance?
>
> When we moved to mm kernel it broke. When it was working - it did not
> show significant effect in quick tests like bonnie.
Fix it.  Find the right test to measure its effect.
>
> > >>What about NFS?
> > >
> > >fixed by nikita recently but not released.  See next snapshot.
> >
> > I'm trying BK pulls. Yesterday it didn't compile, xfs seemed broken.
> > Retrying now, after bk -r get.
> >
> > I wrote a lame script, haven't used it much myself, but it should do
> > something. Like get a BK pull.
> > http://mjt.nysv.org/reiser/getreiser.sh
> > http://mjt.nysv.org/reiser/getreiser.sh.txt
> >
> > >by doing the architectural work....  Also, it requires that 2/3 of the
> > >work (by cost) be done in the US, and the moscow embassy has gone into
> >
> > Have you tried other governments?
> >
> > >total asshole mode and isn't giving work visas to most russians.  Still,
> > >we will try to get work visas in 6 months.  I hope that it being for DoD
> > >work will increase the chance of getting visas.
> >
> > That must not fail.
> > The DoD can't say "please give these guys work visas" just to cover
> > their own interests? Or this has no effect?
> >
> > >It is in the source code..... we have the best commented source code in
> > >the kernel (Andrew Morton is the only other kernel person who comments
> > >code well that I know of).
> >
> > There was also a tangent on the topic of creating daily snapshots.
> >
> > This might make it easier for people to come to terms with Reiser4.
> > People have many reasons to dislike BK, so a daily snapshot patch
> > against some vanilla system may be a good thing.
> >
> > However, having broken code in the BK for days without having time to
> > fix it is not good.
> >
> > So this would require changelogs and some testing on snapshots so that
> > every once in a while one can be labeled "stable as hell"