[v4] reiser4 dream came true... in Solaris9.

[v4] reiser4 dream came true... in Solaris9.

[Nikita Danilov]

Solaris9 supports extended attributes accessible through usual file
system namespace. Support in cp, tap, cpio.

http://docs.sun.com/db/doc/816-0220/6m6nkorp9?a=view

Nikita.

Re: [v4] reiser4 dream came true... in Solaris9.

[Hans Reiser]

Nikita Danilov wrote:

>Solaris9 supports extended attributes accessible through usual file
>system namespace. Support in cp, tap, cpio.
>
>http://docs.sun.com/db/doc/816-0220/6m6nkorp9?a=view
>
>Nikita.
>
>
>  
>
What they did seems like what someone not really understanding what we 
are doing would copy from our implementation if some guy from marketing 
said that they needed it to compete with us.  They don't really move 
attributes into the namespace.  They access them with an API separate 
and additional to that used for files.  The only thing they do get right 
is that they store them as files and directories.  I bet that they don't 
have the underlying infrastructure needed for this to be efficient for 
small attributes though, and I bet the performance sucks.

Sorry for my ungenerous comments.

-- 
Hans

Re: [v4] reiser4 dream came true... in Solaris9.

[Hendrik Visage]

On Mon, Aug 04, 2003 at 05:30:47PM +0400, Hans Reiser wrote:
> small attributes though, and I bet the performance sucks.
> 
> Sorry for my ungenerous comments.

Everybody is entitled to their own opinions ;^)

But thanks for the other info/differences on it.

HEndrik

Re: [v4] reiser4 dream came true... in Solaris9.

[Nikita Danilov]

Hans Reiser writes:
 > Nikita Danilov wrote:
 > 
 > >Solaris9 supports extended attributes accessible through usual file
 > >system namespace. Support in cp, tap, cpio.
 > >
 > >http://docs.sun.com/db/doc/816-0220/6m6nkorp9?a=view
 > >
 > >Nikita.
 > >
 > >
 > >  
 > >
 > What they did seems like what someone not really understanding what we 
 > are doing would copy from our implementation if some guy from marketing 
 > said that they needed it to compete with us.  They don't really move 
 > attributes into the namespace.  They access them with an API separate 
 > and additional to that used for files.  The only thing they do get right 

You missed a bit about runat(1) command:
http://docs.sun.com/db/doc/816-0210/6m6nb7mjs?a=view

About their copying this feature from reiser4: man pages are tagged as
"SunOS 5.9  Last Revised 22 Jun 2001". 

 > is that they store them as files and directories.  I bet that they don't 
 > have the underlying infrastructure needed for this to be efficient for 
 > small attributes though, and I bet the performance sucks.
 > 
 > Sorry for my ungenerous comments.

You should file IP violation scam^H^H^Huit against Sun.

 > 
 > -- 
 > Hans
 > 
 > 

Nikita.

Re: [v4] reiser4 dream came true... in Solaris9.

[Hans Reiser]

Nikita Danilov wrote:

>Hans Reiser writes:
> > Nikita Danilov wrote:
> > 
> > >Solaris9 supports extended attributes accessible through usual file
> > >system namespace. Support in cp, tap, cpio.
> > >
> > >http://docs.sun.com/db/doc/816-0220/6m6nkorp9?a=view
> > >
> > >Nikita.
> > >
> > >
> > >  
> > >
> > What they did seems like what someone not really understanding what we 
> > are doing would copy from our implementation if some guy from marketing 
> > said that they needed it to compete with us.  They don't really move 
> > attributes into the namespace.  They access them with an API separate 
> > and additional to that used for files.  The only thing they do get right 
>
>You missed a bit about runat(1) command:
>http://docs.sun.com/db/doc/816-0210/6m6nb7mjs?a=view
>
Oh, this is so ugly.  You have to do this separately with each directory 
you want to access.

>
>About their copying this feature from reiser4: man pages are tagged as
>"SunOS 5.9  Last Revised 22 Jun 2001". 
>
Well, I don't remember exactly when we started discussing this on lkml 
and on our website, it was maybe 1999?, not sure, but you are probably 
right that it is not copied.  It is a bit obvious as inventions go.  
Well, what they did is better than nothing, but I think it is 
insufficiently thorough.

-- 
Hans