Hallo,
attached find a very conservative patch against your patched
cmd-load-buffer.c which brings the number of calls to realloc()
from n=(final buffer size) to n
tmux.diff
Description: Binary data
At Thu, 18 Feb 2010 23:20:19 +,
Nicholas Marriott wrote:
>
> Hi
>
> Thanks for the diff.
>
Yes, this is quite a more linear patch.
It has one defect, in my opinion, i.e. that it does
buffer-size reallocations, usually this is avoided
by increasing the buffer by a factor 2 rather than increasing
it by 1 byte. That amounts to changing
--
Hallo,
please find enclosed a patch to tmux cmd-load-buffer.c
which reads the buffer from undecidable size streams up to EOF.
I put a predefined limit on the final buffer size.
This, I checked on my currently available machines: netBSD and Linux
and works fine.
Also, I wrote code to read from pipe
xpect.
>
>
> On Wed, Feb 17, 2010 at 01:57:54AM +0100, Fulvio Ciriaco wrote:
> > Please would you explain it better to me?
> > a. does this work on openbsd?
> > b. (mkfifo fifo; echo est > fifo; cat fifo)
> >gives the obvious result and no side effect, h
x27;t return a sane size for FIFOs. You'd have to add
> some debugging to find out what is going on.
>
> On Tue, Feb 16, 2010 at 09:06:05PM +0100, Fulvio Ciriaco wrote:
> > Hallo,
> > I tried the following on different OSes:
> > #tmux setb "echo 1"
> &g
Hallo,
I tried the following on different OSes:
#tmux setb "echo 1"
#mkfifo test
#echo echo 2 > test &
#tmux loadb test
On NetBSD: does nothing
#tmux showb
echo 1
On Linux: empties buffer and creates all sorts of problems
(broken pipe, killed shell)
tmux showb
My questions:
Why does it not wo
