Xref: utzoo alt.folklore.computers:10545 comp.misc:11856
Newsgroups: alt.folklore.computers,comp.misc
Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!sdd.hp.com!think.com!barmar
From: barmar@think.com (Barry Margolin)
Subject: Re: Jargon is costing MONEY. Use diffs.
Message-ID: <1991Mar27.182044.7250@Think.COM>
Sender: news@Think.COM
Organization: Thinking Machines Corporation, Cambridge MA, USA
References: <1307@otello.sublink.org> <1666@tardis.Tymnet.COM> <1991Mar27.162359.20502@hollie.rdg.dec.com>
Date: Wed, 27 Mar 91 18:20:44 GMT

In article <1991Mar27.162359.20502@hollie.rdg.dec.com> law@sievax.enet.dec.com (Mathew Law) writes:
>I don't know what kind of diff you are using, but when I perform the
>same diff (between 2.7.1 and 2.8.1) I get a file of just under 430000
>bytes.  This is less than half the size of a complete repost.  The diff
>between 2.6.1 and 2.7.1 was between 600000 and 700000 bytes - still less
>than a full repost.

Were you doing a context diff?  They include a couple of lines of
surrounding text for each change.  This makes them preferable when posting
changes to files that might have been edited by users, as the context
allows patch programs to find lines that have moved, or determine that the
user has made their own, incompatible changes to a section that also
appears in the diff.  Since reviewers may be making their own edits to the
file, this is a good feature.  However, the extra lines make the diff
bigger; an ordinary diff that looks like:

123c145
< this is the old line
---
> this is the new line

will look like the following in a context diff:

***************
*** 122,124 ****
  This line hasn't changed at all
! this is the old line
  This line is also the same
--- 144,146 ----
  This line hasn't changed at all
! this is the new line
  This line is also the same

In this little example, the size of the diff nearly quadrupled.  It's
probably not so bad in the jargon file, since many of the changes are whole
entries and one or both of the surrounding lines will probably be a blank.
--
Barry Margolin, Thinking Machines Corp.

barmar@think.com
{uunet,harvard}!think!barmar