Jim,
This is indeed news for me. I shall in that case correct my view in a
seperate article on Byte and duly attribute xfs's achievments.
Many thanks for highlighting these issues.
Kind regards
Moshe Bar
www.moelabs.com
> -----Original Message-----
> From: Jim Mostek [mailto:mostek@xxxxxxx]
> Sent: Thursday, June 01, 2000 2:40 AM
> To: Moshe Bar
> Cc: linux-xfs@xxxxxxxxxxx
> Subject: Re: Article in byte.com
>
>
>
> SGI is pouring lots of $$ into getting XFS into the Open Source
> community. There have been as many as 14 internal people working
> on the project. Right now, there are engineers/contractors in Denmark,
> Canada, Minnesota, California, and Australia working on this project.
>
> Part of the XFS port involved making modification to Linux itself.
> Have you heard of pagebuf? We have exchange e-mail/phone
> conferences/... with the ReiserFS and JFS people about joining us
> with the pagebuf work. Pagebuf should be usable by other file systems.
>
> There are some (but few) "external hackers" helping with the code.
> Since we open sourced, about 4-5 patches have come in from people
> outside SGI.
>
> All of the XFS code is now unencumbered. The final source files
> were completed in March.
>
> I don't think we should view the JFS, ReiserFS, XFS code as
> being in competition. These groups should work together to make
> Linux a better OS. Each file system will make its mark. XfS is
> fully multi-threaded and highly scalable. It is also journaled
> and has other capabilities like Direct I/O, extended attributes,
> DMAPI, GRIO, ... Some of these capabilities require changes
> outside the file system and this work should be done with others.
>
> The XFS team is looking forward to working with others and
> will be attending Usenix (Freenix track) and other conferences
> where we should meet and discuss future joint work.
>
> Jim
>
>
> >
> >Dear Jim
> >
> >I was actually aware of an early release of xfs, but I was told
> by sgi that
> >it was still encumbered by licencing restrictions on parts of the code.
> >
> >Additioanlly, it seemed that contrary to IBM, sgi was putting
> very limited
> >resources of its own into the project and was instead almost entirely
> >relying on external hackers to complete the project.
> >
> >Is that true? Can you clarify?
> >
> >Many thanks for writing.
> >
> >Kind regards
> >
> >Moshe Bar
> >
> >
> >
> >> -----Original Message-----
> >> From: Jim Mostek [mailto:mostek@xxxxxxx]
> >> Sent: Wednesday, May 31, 2000 7:20 PM
> >> To: moshe@xxxxxxxxxxx
> >> Cc: linux-xfs@xxxxxxxxxxx
> >> Subject: Article in byte.com
> >>
> >>
> >>
> >> Moshe,
> >>
> >> In your article http://www.byte.com/column/BYT20000524S0007, it
> >> appears you are unaware that SGI has released the XFS code.
> This was done
> >> a few months ago and the announcements appeared in slashdot and other
> >> places.
> >>
> >> Many thousands of people have downloaded XFS and are playing with it.
> >> We have an entire 2.3 kernel tree available for download. (see
> >> http://oss.sgi.com/projects/xfs).
> >>
> >> All file system functionality is there, but there are all kinds
> >> of bugs still in the code. We are also adding more XFS functionality
> >> before first release. The current source even has "delayed" allocation
> >> implemented where each writes do not result in an allocation. This is
> >> pretty rough right now and needs more work, but the main code is there.
> >>
> >> The implementation is not beta or alpha quality, yet. We
> expect to have a
> >> beta quality version in the coming months.
> >>
> >> FYI,
> >>
> >> Jim
> >>
> >
>
|