Re: [PATCH v18 0/1] MR7400: cmd: Allow any key to continue past DIR /P pauses.

[eric pouech (＠epo)]

On Thu Feb 27 16:00:21 2025 +0000, Joe Souza wrote:
> I'll answer #2 first:
> Under Windows, Ctrl-C definitely does break out of a DIR /P pause.
>  Regarding the second part, I completely agree that the proper way to
> handle this would be to wait on both input and the Ctrl-C event.  I have
> tried this.  You may recall in an earlier iteration of my changes for a
> prior merge request, I left some code commented out which I then removed
> after you raised issue with it.  This code was a call to
> WaitForMultipleObjects, waiting on both the input handle and the Ctrl-C
> event.  The problem here which I believe I had stated at the time is
> that with Wine, the input handle is always signaled, even if there is no
> input waiting.  The result was that the wait was always being satisfied
> immediately and DIR /P was not pausing at all.  So that approach was
> abandoned in favor of what we have now.
> #1:
> This code here is based on the previous code for WCMD_Pause.  That code
> was setting console mode flags to 0.  I'm guessing that PAUSE has
> operated this way for many years.  My changes use the new
> WCMD_wait_for_input for both PAUSE and DIR /P, and I did not want to
> break PAUSE behavior.  Further, ENABLE_ECHO_INPUT is on by default, and
> counter to Microsoft's docs for SetConsoleMode, on Wine the character is
> echoed even if ENABLE_LINE_INPUT is disabled.  The end result here is
> that DIR /P was echoing the character that the user pressed, whereas
> Windows does not exhibit that behavior.
> All that said, based on shortcomings in Wine that I discuss above, I
> believe that my changes here are currently the best approach.
> the attached patch (on top of what you sent earlier on) should work as expected:

* the annoying parts: console handles don't support overlapped mode, and input console handles are signaled when there's a pending event... but a pending event could be mouse move, window resize, ctrl key down & up... maybe that's the reason you always check it as signaled
* so the attached patch waits on both console input handle and control c event, but has to discriminate when the console input handle is signaled whether there's an actual character or some other things 
* it's still racy at some places, but given the API I don't see a clean way to do it
* [wait-con.patch](/uploads/a537c394d4b98165b791a7979ad3f8ca/wait-con.patch)

-- 
https://gitlab.winehq.org/wine/wine/-/merge_requests/7400#note_96140