Re: glib outstanding stuff

[David Benson <daveb ffem org>]

except if the file is a fifo (or an entry in /proc),
so probably a fallback to havoc's original implemention
would be a good idea.

- dave

On 4 Oct 2000, Havoc Pennington wrote:

> 
> Joel Becker <jlbec evilplan org> writes: 
> > 	If the file is too big, the initial g_new() will fail, so there
> > is no crash.  In addition, this saves all the resizings of the g_string.
> > As we are willing to read as large a chunk as the kernel will give us,
> > we don't need buffering ala stdio.
> > 	This is what I have in my personal convenience libs (modulo
> > error checking, of course).  If there's a good reason for your version,
> > I'd like to know so I can change. :-)
> > 
> 
> Yes, your version looks better to me (modulo the fact that g_new()
> does indeed crash if it fails - but we can just replace it with
> g_try_malloc() and it will be fine).
> 
> Havoc