Date:	Tue, 30 Jan 1996 13:06:43 +0000 (GMT)
From:	Andy Molyneux <molyneua AT cs DOT man DOT ac DOT uk>
To:	Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
Cc:	djgpp AT delorie DOT com
Subject:	Re: Compiler abort!

On Mon, 29 Jan 1996, Eli Zaretskii wrote:

> This bug doesn't happen in GCC 2.7.2 (in the next beta), so don't bother 
> reporting it to FSF.
> 
Fair enough. I wasn't going to, truth be told!

> Nothing's really wrong, except that it's a no-op: it changes structures
> passed by value, so the calling program won't ever see the change.  My
> guess is GCC sees this and tries to optimize out most of the code, which
> makes some of its parse tree invalid later.
>

I think you've misunderstood me slightly... the sample I posted was but a
fragment of a much larger function which had the same catastrophic effect.
Rest assured that the original function DID use the structures! In case
you're interested (probably not), the function's purpose was to draw a
clipped, filled triangle on the screen. Anyway, I'm pretty sure it's got
nothing to do with optimizing away unused values. 

Weird, isn't it?

- Andy

| Andrew Molyneux            Internet: molyneua AT cs DOT man DOT ac DOT uk
| University of Manchester   Tel     : (+44)161-485-3291
| REAL programmers write  a=b!=4?b%2?c:d:e  without even
| thinking about it :-)

- Raw text -