Re: malloc and alignment

[David.Kastrup@t-online.de (David Kastrup)]

"Stephen J. Turnbull" <stephen@xemacs.org> writes:

> >>>>> "Stefan" == Stefan Monnier <monnier+gnu/emacs@rum.cs.yale.edu> writes:
> 
>     Stefan> but his Emacs process went up to 2.1GB (Mule begins by
>     Stefan> allocating a destination area of 1510MB because latin-1
>     Stefan> has a max-expansion factor of 2),
> 
> In XEmacs, that's a hypothesis, IIRC, not yet proved; it could just be
> a bug in the allocation routines.
> 
>     Stefan> so I think that a 512MB limit due to integers might not be
>     Stefan> that bad, since bumping it to 1GB will just replace this
>     Stefan> limit with others.
> 
> FWIW when I proposed swapping 31-bit integers and 30-bit characters so
> that we could have a proper native representation of ISO 10646 UCS-4,
> it was vetoed on the grounds that the vetoer's coworker edited a
> particular (ASCII-only) > 512MB logfile every day.

Sounds like transparent big integers might be needed at some time.
There is always going to be someone to cry foul because one bit less
does not cut it for him: after all, it was necessary to remove this
limitation from file systems, too.


-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum