Re: process output has become a bit random...

[David Kastrup <dak@gnu.org>]

storm@cua.dk (Kim F. Storm) writes:

> Peter Heslin <usenet@heslin.eclipse.co.uk> writes:
> 
> > On 2004-07-28, David Kastrup <dak@gnu.org> wrote:
> > >  Has anybody else experienced anything weird in connection with process
> > >  output recently?
> > 
> > When running M-x grep recently, I got the output duplicated
> > several times, which happened on a couple of random occasions.  At
> > the time I thought it might have been related to the super-slow
> > grep output bug, but it sounds pretty close to what you describe.
> 
> I made a change on 2004-06-07 which increased the read buffer from
> 1k to 4k.  I don't see how that could provoke duplicate buffer
> output, though.

Well, something's rotten in the state of Denmark, anyway.  If you take
a look at the buffer sizes, you'll see that chars is allocated with a
size of carryover+readmax, but it is only ever filled with carryover
old and (readmax-carryover) new characters, for a total of merely
readmax characters.  Consequently, I had at one time patched down
either the buffer size or increased the read sizes (don't remember
which it was), but that was later found to cause segmentation faults.
So the change was reverted, but never explained.

I have no clue whether this might be related.

> Another change which may be relevant is this
> 
> 2004-06-11  Kenichi Handa  <handa@m17n.org>
> 
> 	* coding.c (decode_coding_string): Check CODING_FINISH_INTERRUPT.
> 
> Process output handling calls decode_coding_string in such a way that
> it may not decode an entire string (leaving further decoding to the
> next call).  If there is some error in that logic, I think text could
> be duplicated.

That's a possibility, yes.  Another remote one is that it may coincide
with myself changing from gcc-3.3.x to gcc-3.4.1.

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum