Path: utzoo!attcan!uunet!lll-winken!lll-tis!ames!mailrus!tut.cis.ohio-state.edu!rutgers!mit-eddie!uw-beaver!ssc-vax!bcsaic!paula
From: paula@bcsaic.UUCP (Paul Allen)
Newsgroups: comp.os.minix
Subject: Re: printer driver , 1.3c, checksums (really patch vs. fix)
Message-ID: <7523@bcsaic.UUCP>
Date: 9 Sep 88 05:33:24 GMT
References: <630@faui44.informatik.uni-erlangen.de>
Reply-To: paula@bcsaic.UUCP (Paul Allen)
Organization: Boeing Computer Services AI Center, Seattle
Lines: 46

In article <630@faui44.informatik.uni-erlangen.de> rtregn@immd3.informatik.uni-erlangen.de (Robert Regn) writes:
>
>In <850@ast.cs.vu.nl> ast@cs.vu.nl (Andy Tanenbaum) writes: 
>>... and make up the final, definitive, never-again-to-changed V1.3,
>But please send no update vs. 1.3b but vs. 1.3a (or 1.2).
>Many people can so skip a step and spare a lot of work.
>(Updating kernel, lib and commands requires about 30 hours on an AT)

I agree that it would be less work if the final 1.3c diffs were relative
to 1.2, but is MINIX really that slow?  I keep my canonical reference
copies of 1.1, 1.2, and 1.3* on a Sun.  The major time-consumer when
generating 1.2 was in assembling the *true* set of 1.2 diffs!  Once I
had settled on which diffs really defined 1.2, a script something like
the following applied them in just a few minutes:

for n in *.dif
do
	patch `basename $n .dif` $n
done

Then all I had to do was go through the .rejects files and fix up the
hunks that didn't apply properly.  I can believe that *compiling* the
whole mess would take a while!  I have not been able to get all the way
through a complete make of everything on my Sun IPC.  MINIX won't talk
to the IPC's hard disk because it is really emulated by a file in the
Sun file system.  Running a make on the IPC's floppies usually results
in a diskette error after about a half hour of clunking away.  I suspect
some sort of timing glitch in the floppy driver, but haven't spent much
time on it.

>>Discussion on the subject of cdiff/patch vs diff/fix should be posted
>I like diff/fix. The cdiffs are MUCH to big. The context is not necessary
>because many people don't look at the diff's. People with own fixes should
>have official versions too.

Patch is much more robust than fix.  I attempted to use fix to apply 1.2
and gave it up as unuseable.  With due respect to the people who wrote
and enhanced the fix program, patch is better.  I am willing to pay the 
space penalty of context diffs in order to have the context around when 
a diff collides with a local hack and I have to apply it by hand.

'nuff for now...
Paul
-- 
------------------------------------------------------------------------
Paul L. Allen                       | paula@boeing.com
Boeing Advanced Technology Center   | ...!uw-beaver!ssc-vax!bcsaic!paula