From owner-freebsd-hackers  Wed Feb  7 03:43:15 1996
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id DAA22439
          for hackers-outgoing; Wed, 7 Feb 1996 03:43:15 -0800 (PST)
Received: from ref.tfs.com (ref.tfs.com [140.145.254.251])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id DAA22426
          Wed, 7 Feb 1996 03:42:35 -0800 (PST)
Received: (from julian@localhost) by ref.tfs.com (8.6.12/8.6.12) id DAA00568; Wed, 7 Feb 1996 03:42:29 -0800
Date: Wed, 7 Feb 1996 03:42:29 -0800
From: Julian Elischer <julian@ref.tfs.com>
Message-Id: <199602071142.DAA00568@ref.tfs.com>
To: hackers@freebsd.org
Subject: corrected context diffs of terries patches
Cc: terry@freebsd.org
Sender: owner-hackers@freebsd.org
Precedence: bulk


I've been through his patches.
I would really like someone ELSE
so go through them too. I didn't see any problems, and I did like
the restructuring of who frees what with regards to the namei name buffer
it was definitly broken before..

I've placed a file in 
ftp::/freefall.cdrom.com:/incoming/context-diffs-terry
with a slightly fixed up version of his diffs. (fixed a spammed file)


comments.
The first HALF (actually 60%) is MOSTLY cosmetic changes
terry re-wrote almost all of vfs_syscalls.c
so that all functions have a single exit point.
My guess is that he plans to be able to do somethig at this exit point 
at some point in the future. (?)

I am in two minds about this..
1/ I think it's slightly cleaner code
2/ it might be a few cycles slower in places
3/ It makes it harder to merge in 4.4lite2 patches
4/ It's not ALWAYS easier to read, and is often harder to read
because you have to find the end of the nested if'else'if sets
to figure out where the code goes, while 'return EINVAL;'
is pretty bloody obvious.
5/ It's easier to have leaks of things with many exit points 
6/ it obfuscates any REAL changes he's made in the area and might
intruduce new bugs for little gain.

I'm kinda against these cosmetic patches.. unless terry can give a good reason.
Terry, if you could extract the REAL changes from this stuff and submit that
I'd be pretty happy with the patch.

the last 40% of the context diff is real changes to do with the
re-assigning of responsibility for freeing the name buffer in the namei
code. This is a good idea, though I haven't checked 100% of the change yet.
I think this part should be applied..

Any one Else care to look at it and comment?
especially on the last 40%?
If I can get terry to undo the cosmetic changes and submit them totally
separatly the first 60% of the patch would shrink to about 10%..

julian