From owner-freebsd-current  Wed Apr  5 22:28:04 1995
Return-Path: current-owner
Received: (from majordom@localhost)
          by freefall.cdrom.com (8.6.10/8.6.6) id WAA10739
          for current-outgoing; Wed, 5 Apr 1995 22:28:04 -0700
Received: from mail.barrnet.net (mail.BARRNET.NET [131.119.246.7])
          by freefall.cdrom.com (8.6.10/8.6.6) with ESMTP id WAA10732
          for <current@FreeBSD.org>; Wed, 5 Apr 1995 22:28:03 -0700
Received: from estienne.cs.berkeley.edu (estienne.CS.Berkeley.EDU [128.32.42.147]) by mail.barrnet.net (8.6.10/MAIL-RELAY-LEN) with ESMTP id WAA01436 for <current@FreeBSD.org>; Wed, 5 Apr 1995 22:24:50 -0700
Received: from localhost (localhost [127.0.0.1]) by estienne.cs.berkeley.edu (8.6.11/8.6.9) with SMTP id WAA21470; Wed, 5 Apr 1995 22:26:01 -0700
Message-Id: <199504060526.WAA21470@estienne.cs.berkeley.edu>
X-Authentication-Warning: estienne.cs.berkeley.edu: Host localhost didn't use HELO protocol
To: Kai.Vorma@hut.fi
cc: current@FreeBSD.org
Subject: Re: patch for sup 
In-reply-to: Your message of "Wed, 05 Apr 1995 12:15:45 +0300."
             <199504050915.MAA10911@vinkku.hut.fi> 
Date: Wed, 05 Apr 1995 22:26:00 -0700
From: "Justin T. Gibbs" <gibbs@estienne.CS.Berkeley.EDU>
Sender: current-owner@FreeBSD.org
Precedence: bulk

>Here is a simple patch for sup that makes it possible to have local
>modifications in your source tree so that sup won't overwrite them. The
>idea is simple (stolen from this mailing-list, I think :)
>
>  If sup is updating file foo.c and there exists file foo.c#sup it
>  will update the latter file an leave the former untouched.
>
>This patch should work with regular files and perhaps with links but
>you _cannot_ create shadow directory <dir>#foo. I have not tested this
>too much (I did it yesterday evening) but I has worked so far just
>fine :-)
>
>Btw, why is freefall's supserver so abysmally slow? Is it because
>supservers uses fork() like almost all badly designed network
>servers or something else? If the problem is not fork() then I'll try
>to see if I can make it faster.
>
>..vode
>
>PS. This patch is agains sup.tar.gz from ports-collection.

Can you rewrite this so that it doesn't double the number of stat calls
needed to do an upgrade?  I think you should be doing this check only
after a file has been deemed an upgrade target and the test should be made
conditionally on a an option to SUP.  The test could be done much more
simply by doing the stat on upgrade_target.suffix and then if it exists,
change the name of the target in the tnode and goto the top of the
"upgrade" test cases and loop through them a second time.  That way we 
penalize the special case and leave the common case alone.

>+ 
>+ 
>+ char *SUFFIX = "#sup";
>+ 
>+ int tdostat(ls, path, sbuf)
>+      int ls;
>+      char *path;
>+      struct stat *sbuf;
>+ {
>+   int x, l = strlen(path);
>+ 
>+   strcpy(rname, path);
>+   if (l + strlen(SUFFIX) < sizeof(rname)) {
>+     strcat(rname, SUFFIX);
>+     if ((x = ls ? lstat(rname, sbuf): stat(rname, sbuf)) != -1) {
>+       if (sbuf->st_mode&S_IFMT == S_IFDIR)
>+ 	goaway("%s must not be a directory", rname);
>+       return x;
>+     }
>+     if (errno != ENOENT)
>+       goaway ("%sstat for %s failed (%d)", ls ? "tl" : "t", rname, errno);
>+     rname[l] = '\0';
>+   }
>+   return (ls ? lstat(rname, sbuf) : stat(rname, sbuf));
>+ }
>+ 

--
Justin T. Gibbs
==============================================
TCS Instructional Group - Programmer/Analyst 1
  Cory | Po | Danube | Volga | Parker | Torus
==============================================