Re: Next pretest

[Juanma Barranquero]

On Tue, Jan 26, 2010 at 19:09, Eli Zaretskii <address@hidden> wrote:

> Probably not, since it triggers an assertion.

True, but sometimes assertions grow stale, and nobody notices for a
long time because apparently not many developers build with assertions
enabled.

> Does this happen with 12XX codepages as well?

  C:\> chcp 1252
  C:\> emacs -Q -nw
  ñ

produces the same effect.

> This is a non-optimized build, I presume, right?  We aren't chasing
> wild geese here, right?

"--with-gcc (4.4) --no-opt --cflags -DENABLE_CHECKING=1 -DXASSERTS=1
-IC:/emacs/build/include -fno-crossjumping"

I'm using the Twilight Dragon Media MinGW build, but I'm reasonably
sure it happened too with stock MinGW gcc 3.X, so it's not likely a
compiler bug.

> Can you see what event is received w32term.c:w32_read_socket in this
> case?  Is that function at all called in the -nw session?  If not, try
> w32inevt.c:key_event.  In the latter, it would be interesting to see
> what was event->wVirtualKeyCode (around line 426).

I cannot debug this right now, but I'll try tonight.

> Also, how did you type these non-ASCII characters on your keyboard?
> Do you have keys for them, or do you need to use AltGr or some such?

They have their own keys (I'm using a Spanish keyboard). But AltGr-4 +
n (which produces ñ) triggers the assertion too.

    Juanma