Re: [ANNOUNCEMENT] CALL FOR TESTING: Cygwin 1.7.12

[Thomas Wolff <towo@towo.net>]

Am 02.04.2012 19:05, schrieb Christopher Faylor:
> On Mon, Apr 02, 2012 at 10:07:21AM +0200, Corinna Vinschen wrote:
>> On Mar 31 14:01, Thomas Wolff wrote:
>>> Am 30.03.2012 13:33, schrieb Corinna Vinschen:
>>>> ...
>>>> - Fix a bug in controlling tty handling when duplicating a console descriptor.
>>> Hi, when I read this, I wondered whether this bug could be fixed:
>>> http://sourceware.org/bugzilla/show_bug.cgi?id=513
>>> but it isn't.
>>> I had later found (and reported to cygwin-developers) that the
>>> terminal response to terminal queries (like cursor position request
>>> or device attribute request) is placed in the wrong buffer because
>>> there are different objects (descriptors?) for stdin and stdout, so
>>> the response would stay in Nirvana and never reach the application.
>>> (And it used to work in CYGWIN=tty mode because curiously, in that
>>> case, the console handles would have been shared for stdin and
>>> stdout.)
>>> Since you seem to have just looked at this area of the console code,
>>> could you give it check?
>> I cant't find your later mails on the subject in the cygwin-developers
>> list archive, but I vaguley remember that this was discussed at one
>> point and nobody had an idea how to implement it without turning the
>> code upside down.  You're familiar with the fhandler_console code as
>> well, so maybe you have an idea now how to implement this?
Not so familiar, actually, except with some very specific parts now. But 
I tried something, see my patch to cygwin-patches and my first comments 
there.

> I don't see how we could accommodate this without switching the console
> to "uncooked" mode.  And, as you say, that would have massive effects
> on the console code.  I don't think it's worth that much change for such
> a relatively unused feature.  Especially since it would make console I/O
> slower.
I don't see how it should make anything slower. It may not be worth it 
after all, but there could be a correlation with select() problems 
(input being delayed and not recognized immediately).
> That said, I think the bug should be closed WONT_FIX.
As I indicated on cygwin-patches, it may be worth a check anyway.

Thomas

--
Problem reports:       http://cygwin.com/problems.html
FAQ:                   http://cygwin.com/faq/
Documentation:         http://cygwin.com/docs.html
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple