From:	"jansb000" <jansb000 AT wxs DOT nl>
Newsgroups:	comp.os.msdos.djgpp
Subject:	Re: Need help with very nasty bug
Date:	Tue, 24 Jul 2001 17:53:40 +0200
Organization:	Planet Internet
Lines:	28
Message-ID:	<9jk51v$1jjmp$1@reader02.wxs.nl>
References:	<9ii9h3$r9om$1 AT reader03 DOT wxs DOT nl> <9iihco$bmm$1 AT nets3 DOT rz DOT RWTH-Aachen DOT DE> <3b4fbf4d DOT 165926315 AT news DOT primus DOT ca>
NNTP-Posting-Host:	ipc3797b3d.dial.wxs.nl
X-Trace:	reader02.wxs.nl 995989375 1691353 195.121.123.61 (24 Jul 2001 15:42:55 GMT)
X-Complaints-To:	abuse AT wxs DOT nl
NNTP-Posting-Date:	24 Jul 2001 15:42:55 GMT
X-Priority:	3
X-MSMail-Priority:	Normal
X-Newsreader:	Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE:	Produced By Microsoft MimeOLE V5.50.4522.1200
To:	djgpp AT delorie DOT com
DJ-Gateway:	from newsgroup comp.os.msdos.djgpp
Reply-To:	djgpp AT delorie DOT com

> Actually, much can be deduced without more info. All he changed was
> the object layout and the program broke. Ergo, its behavior has a
> dependency on object layout, which is bad. Most probably, this came
> from one of three causes:
>
> * A bug involving uninitialized memory or pointer wildness; the
>   program happened to always work before the spr was added perhaps
>   because a padding zero was there before, or another field.

Unlikely since I reshuffeled (is this English ?) the attributes with exactly
the same result.

> * The code was unwisely depending on fields of the object being at
>   certain hardcoded offsets in bytes, rather than using offsetof.

No. I never use or depend on hardcoded offsets.

> * The code was unwisely writing out and reading the objects
>   byte-by-byte to and from binary files, and so the object format
>   changed when the spr field was added, making the program choke on
>   earlier versions' files.

Bingo!, Indeed my testprogram reads a binary file (containing this class)
written with an older (unmodified) version of the program.

Thanx for the tips!

- Raw text -