Subject: Re: SGI will freely license its XFS
To: Miles Nordin <carton@Ivy.NET>
From: Dave McGuire <>
List: current-users
Date: 05/20/1999 19:03:24
  Don't hold back, man.  Tell us what you *really* think.

  I can just FEEL the love...

      -Dave McGuire

On Thu, 20 May 1999, Miles Nordin wrote:
>On Thu, 20 May 1999, Eric S. Hvozda wrote:
>> Is LFS ready for prime time?
>as opposed to, a single-architecture in-house corporate JFS implementation
>written for a highly-customized version SVR4 offered under hazy licensing
>terms and proposed for integration into the BSD-licensed kernel?
>Given that Linus was the man who ``blessed'' Slovansen's sound drivers,
>let Ledford's ``egregarious'' aic7xxx hacks into the so-called stable
>kernel, and agreed with the idea of a ``Network Block Device Driver,''
>this XFS thing has the potential to be very, very amusing. 
>Remember, we're talking about the UNIX that threw out a general-purpose
>message-passing kernel-to-userland interface for the stated purpose of
>making an include file 40% shorter and modularizing UNIX-domain sockets.
>Include-file-bloat is indeed a major performance problem, and UNIX sockets
>use almost 10k of kernel memory.  Who uses UNIX sockets, anyway? X?
>syslogd? Best that the 10k be freed.
>My favorite quite from the press release:
> ``Without a journaling file system, rebooting a computer after a crash
>   can be extremely time-consuming, a critical issue as companies look for
>   computer systems that are available to users around the clock.''
>As someone who's seen what ext2fs does to the filesystem on a dying disk,
>I'd just like to take a moment of everyone's time to announce my gratitude
>to the Linux community, on behalf of myself and my users:  ``hear, hear!
>What High Availability really needs is shorter fsck times!  I'm tired of
>tweaking rc on my Status to skip fsck.  fsck should be a kernel module.  
>All hail, Linux XFS.  Assemble a team.  Work on fsckmod will begin
>On 20 May 1999, Oleg Polyanski wrote:
>> lfs is an example of excellent research work but in real life it is
>> unusable.
>since it is research, I think the correct term in this case is
>``unfinished.''  In any case, it can't possibly be worse than the VxFS in
>HPUX 10.20.
>If it's not clear, I'll take the Pepsi Challenge with that XFS stuff any
>day.  If just one company would turn the license fees for 5 copies of Irix
>into an LFS research grant, I bet we'd be through this small mess.
>sick of white papers,
>Miles Nordin / 1-888-857-2723
>555 Bryant Street PMB 182 / Palo Alto, CA 94301-1700