Re: tlug: ccp

["Manuel M. T. Chakravarty" <chak@example.com>]

Chris Sekiya <chris@example.com> wrote,

> On Tue, 15 Dec 1998, Manuel M. T. Chakravarty wrote:
> 
> > BTW, I wouldn't trust egcs as a replacement of gcc in the first place.  
> > I am pretty sure that it breaks sometimes, ie, generates bad code.
> 
> It can.  Up until 1.1.1 there were gcse problems in targets that didn't
> have a plethora of registers (i.e., Intel).  I've had pretty good
> experiences with egcs.

Right, the problem seems to be a register allocation problem.

> > I tried to use egcs instead of gcc and got some pretty wired error
> > messages about registers which should have not been spilled etc.
> 
> Not surprising.  gcc was pretty lenient about asm, would let you do things
> that you really shouldn't be doing ;)
> 
> egcs is much stricter.

I see.  It is actually my first experience with egcs.  But
why should egcs care about what I do with asm?  I mean, the
whole point of asm is to have a backdoor to the raw
processor machinery - which can be very convenient;
especially, if you use C as a kind of portable high-level
assembler for another compiler.

> Have you talked to the egcs guys about this?

No - I think the problem should be much better characterized
before they can make any use out of it.  Unfortunately, the
C code generated by the high-level compiler is almost
incomprehensible for mere humans...

Manuel
------------------------------------------------------------------
Next Nomikai: 15 January 1999, 19:30 Tengu TokyoEkiMae 03-3275-3691
Next Technical Meeting: 13 February, 12:30               Place: TBD
------------------------------------------------------------------
more info: http://tlug.linux.or.jp                     Sponsor: PHT