Re: GNT, and breakage in windows

[Sadrul H Chowdhury]

On 8/10/06, Sadrul H Chowdhury <address@hidden> wrote:

On 8/10/06, Thomas Dickey < address@hidden> wrote:

On Thu, 10 Aug 2006, Sadrul H Chowdhury wrote:

> On 8/10/06, Thomas Dickey <address@hidden> wrote:
>>
>> On Thu, 10 Aug 2006, Sadrul H Chowdhury wrote:
>>
>> > I have just tried two workarounds to fix this: I put in a "sleep(1);"
>> after
>> > each update_panels/doupdate pair. This didn't work. Then I tried to
>> refresh
>> > the screen every second (by using g_timeout_add). This also didn't work.
>> Is
>> > there any other way I can slow things down so that ncurses can be done
>> > updating the screen before it starts doing so again?
>>
>> There's no simple way - essentially what you'd have to do would be to
>> detect if a refresh was incomplete and tell your application to combine
>> things, e.g., make a window move by two columns at a time rather than by
>> one (but of course the input would still "really" be one column).
>>
>> There's no function defined for this, but one could be written, e.g.,
>> checking if there are pending changes to a window after a refresh or
>> doupdate is completed.
>
>
> I have been unable to find a way of doing this. Do I need to play with the
> internals of WINDOW or some other structure?

yes - it would be in struct ldat (which is declared in curses.priv.h).

I might add it as an extension...

That'd be awesome!

Hi. I just removed the following line of code:

 g_io_channel_set_flags(channel, G_IO_FLAG_NONBLOCK, NULL );

where 'channel' is a GIOChannel for stdin. And the scrambleness went away! In hindsight, I suppose I should've detected this before, when I was pointed to 'typeahead'. Thanks for the inputs. This really excites me! :)

Sadrul