|
From: | Aaron Bentley |
Subject: | Re: [Gnu-arch-users] multiple committers, again |
Date: | Thu, 26 Aug 2004 07:40:17 -0400 |
User-agent: | Mozilla Thunderbird 0.5 (X11/20040306) |
address@hidden wrote:
On Wed, Aug 25, 2004 at 08:38:49PM -0400, Aaron Bentley wrote:c-b-v-b0.cpio, c-b-v-p1.cpio, ... c-b-v-cacherev14.cpio, c-b-v-p15.cpio ...Hmm. I think what you're proposing there would be the first non-pfs archive.Oh is that what I'm doing? :-)
Err. Pseudo-File-System. http, ftp, sftp and local archive access are all implemented as pseudo-filesystems.
I'm very probably too ignorant of pfs to really say, but I would have any write commands just write to the tar, then send the whole tar at the close of the pfs session. (is there such a thing?)
Yeah, "the close of the pfs session" exists. But that approach will work very poorly with concurrent access. Two people could make contradictory changes at the same time, then send the whole tar. The current systems have good locking systems to prevent that kind of situation.
Aaron
[Prev in Thread] | Current Thread | [Next in Thread] |