Re: lib_twait.c

[Thomas Dickey]

On Tue, Aug 06, 2002 at 06:22:29AM +0200, Philippe Blain wrote:
> >From Philippe Blain, Bordeaux, FRANCE - <address@hidden>
> My very old computer: P133 - 8,4 Go - 32 Mo Red Hat Linux 7.0
> I have a dream (P-IV 40Go Geforce4 flat screen...)

I have a P-III, P-II and P-I, but the motivation for the first is to allow me
to work from home at times (keeping my day-job lab going is a lot of work...).
 
> To maintainers of 'ncurses'.(and to Mr Dickey)
> Subject: Corrections for ncurses-5.2-20020803+
> 
> Here are some problems I found:
> 
> ----------------------------------------------------------------------------
> ----
> Thanks for your response about lib_getch.c. I'm going to investigate more
> about
> that if I can.

ok (it does look odd to have wgetnstr called from wgetch, but does not seem
to be avoidable - though I should revisit that for the wide-character case).

> ----------------------------------------------------------------------------
> ----
> File : ncurses/tty/lib_twait.c
> 
> Function : _nc_timed_wait(int mode, int milliseconds, int *timeleft)
> 
> OK, infinite timeout seems good now.

yes - my changelog wasn't good enough to see that clearly, but the bug was
introduced to work around some limitations of the BeOS implementation.  So
I moved the checks into that ifdef (and it will have to be re-tested).
 
> And then testing as "if (_nc_timed_wait(...) > 0)" in lib_getch.c,
> lib_mouse.c.
> 
> This would permit to test errno==EINTR in the mouse code for lib_getch.c

yes, that probably would be an improvement.

-- 
Thomas E. Dickey <address@hidden>
http://invisible-island.net
ftp://invisible-island.net