To: new-httpd@hyperreal.com
Subject: Re: Style options with indent 
In-reply-to: Your message of "Tue, 16 Jul 1996 09:39:49 BST."
             <199607160839.JAA00528@tees>
Date: Tue, 16 Jul 1996 22:07:22 -0700
From: "Roy T. Fielding" <fielding@liege.ICS.UCI.EDU>
Message-ID: <9607162207.aa18346@paris.ics.uci.edu>
Sender: owner-new-httpd@apache.org
Precedence: bulk
Reply-To: new-httpd@hyperreal.com

> Well, I've been looking at the enhancements made to newer versions of the
> cvs scripts and having cvs run indent is not actually much of a problem
> anymore. We could therefore have cvs reformat code before it gets
> committed transparently as far as the committer is concerned.

Actually, I wouldn't recommend that.  As much as I like indent for
cleaning up old code, there are still a lot of "whitespace arrangement"
(particularly around comments and blocks of code) that are better
refined by a human perceptive of the context, rather than by a program.
I am also paranoid enough not to trust indent without double-checking
the results and recompiling -- I have never seen it err, but I am paranoid.

Once the code is in a consistent style, I think it will stay that way for
at least a couple major releases -- someone just has to remember to run
indent (or just use the Apache style) before committing a new module,
and before major cleanups.

One thing to note -- if someone plans to reformat an existing file,
please do the reformatting first (without any semantic changes) and
commit the reformatted code before making whatever semantic changes
are intended.  That way we can differentiate between bugs introduced
by reformatting (large diffs, but there should be no bugs) and bugs
introduced by other changes.

.....Roy