Path: utzoo!utgpu!watmath!clyde!att!pacbell!ames!mailrus!bbn!apple!well!shf
From: shf@well.UUCP (Stuart H. Ferguson)
Newsgroups: comp.sys.amiga
Subject: Re: How does Newzap work?
Keywords: newzap flakey weird is it me?
Message-ID: <10307@well.UUCP>
Date: 10 Jan 89 21:17:27 GMT
References: <5079@phoenix.Princeton.EDU> <5121@phoenix.Princeton.EDU> <10234@well.UUCP> <5650@cbmvax.UUCP>
Reply-To: shf@well.UUCP (Stuart H. Ferguson)
Organization: The Blue Planet
Lines: 30

+-- jesup@cbmvax.UUCP (Randell Jesup) writes:
| In article <10234@well.UUCP> shf@well.UUCP (Stuart H. Ferguson) writes:
| >I discovered when working on the IFF library that VDK: has a problem with
| >seeking.  It doesn't correctly return the new position after each seek, so
| >if NewZAP was expecting this information it would probably go berserk.
| 	Seek() is documented as returning the position in the file BEFORE the
| Seek was done.  Could you be more specific?

Silly me.  Of course, Seek() is supposed to return the old position, not
the new position.  I was doing some rather rigorous checking on seeks,
something like (from memory):

	oldpos = Seek (stream, offset, CURRENT_POSITION);
	newpos = Seek (stream, 0L, CURRENT_POSITION);
	if (newpos - oldpos != offset)
		ERROR;

Doing this on CBM provided devices, like df0: and ram:, did the right thing,
but if I did this on vdk: I would consistently get seek errors.  I changed
the code to be less picky:

	oldpos = Seek (stream, offset, CURRENT_POSITION);
	if (oldpos == -1)
		ERROR;

and vdk: then worked correctly.  I don't know exactly what vdk: is doing,
but it's return value seems to be wrong.
-- 
		Stuart Ferguson		(shf@well.UUCP)
		Action by HAVOC