🚀 go-pugleaf

In article <9311@bloom-beacon.MIT.EDU> tada@athena.mit.edu (Michael Zehr) writes:
<Has anyone experienced the following problem with lattice C? (5.02)
<
<variable = &foo.bar.arry[n];
<printf("%d %d\n", (int) variable, (int) &foo.bar.arry[n]);
<
<...yielding two different values!?!
<
<i've noticed it twice in programs, and so far, it appears that if n=0,
<the assignment is correct, but the evaluation in printf is wrong, and
<if n=1, it's the reverse.
<
<does anyone know of cases in which lattice C screws up structure
<addresses and/or offsets?  is this why there's no option to see the 
<assembly code produced by the compiler? (1/2 :-) )

There's a known bug in 5 that causes offsets to be sign-extended.
This would cause the address to be mis-evaluated if it were more than
32K from the base of the structure.

I found this by checking the assembler ouptut via cpr (actually, it
took noticing that instructions changed between startup & execution,
and running with a watch break on them....)

<what do people recommend for a C compiler?  i need one that does a 
<fairly decent job of optimizing, but i'd also like it to be correct...

From what I've seen, Lattice seems to have fewer bugs/release than
Manx. Neither is perfect, though both do better than many people you'd
wouldn't expect it of.

I've heard rumors of a port of gcc, with unhappy results (RMS likes
using stack space).

	<mike

--
The Sword of Damocles is hanging over my head		     Mike Meyer
And I've got a feeling someone's gonna be cuttin' the thread mwm@berkeley.edu
Oh -- woe is me, My life is a misery			     ucbvax!mwm
And all I can see is I'm on the start of a pretty big downer mwm@ucbjade.BITNET