On Mon, 3 Mar 2008 15:17:06 +0100
"Bert Wesarg" <[EMAIL PROTECTED]> wrote:
> Hi,
>
> On Mon, Mar 3, 2008 at 2:58 PM, AVKuznetsov <[EMAIL PROTECTED]> wrote:
> > Hi,
> >
> > I made a macro for word completion via local dict server
> > and tried its stability for huge number of completions.
> > Following bugs were observed.
> >
> > a) The macro uses calltip to show a list of completions.
> > In a long list (~5000 lines) lines superimposes each other.
> > Seems, overflowed "line counter" drops to zero
> > and new lines are "printed" over previous ones.
> 5000 lines! what do you expect? Ok a scrollable calltip would be nice,
> I have tried this several times, but gived up because I didn't get the
> scrollbars in automatic mode, ie show them only if needed.
>
> If any this is a motif bug.
>
> >
> >
> > b) Creation of long lists completely hangs the system. The
> > cause for the hanging is described as follows.
> >
> > In the following simple
> > string = ""
> > for(i=0; i<N ; i++)
> > string = string "\n" i
> >
> > each iteration eats memory since new string is created.
> > On my system (1.2G memory + 800M swap) a top limit of iterations
> > is 19800 and string of length of 107K is created. Both memory and
> > swap are full and AND THEY ARE NOT FREED WHEN MACRO IS COMPLETED!
> > Simple estimation shows that the memory occupation corresponds to
> > number of created interim strings.
> You can have this behavior much easier without NEdit:
>
> int
> main(int ac, char *av[])
> {
> for (unsigned long i; i < ULONG_MAX; i++)
> void *p = malloc(i);
> return 0;
> }
>
> So what do you expect, should NEdit limit the amount of usable memory?
> BTW: does NEdit crash?
>
> I would call both points programmer errors.
>
Bert, thank you for the explanation.
Sorry, I am not a programmer.
Prof. Alexey V. Kuznetsov, Dr. Sci,
... etc.
P.S.
I prevent my macro from both calltip destruction and memory
overflow, but I am not sure that described overflows
are correct for the NEdit.
--
NEdit Develop mailing list - [email protected]
http://www.nedit.org/mailman/listinfo/develop
