Re: [openMSX-devel] Testing RC1

[Bas W. <b.w...@ph...>]

On Wed, May 26, 2004 at 02:49:23AM +0200, Maarten ter Huurne wrote:
> On Tuesday 25 May 2004 10:20, Manuel Bilderbeek wrote:
> > I know this, because at the first=20
> > run of the new binary, I got this after closing Catapult while openMSX
> > was running:
> > catapult: src/openMSXController.cpp:268: wxString
> > openMSXController::GetPendingCommand(): Assertion `!m_commands.empty()'
> > failed.
> > The problem is that this bug isn't easily reproducible...
>=20
> We could try reviewing the code to see if we can find a scenario in which=
 this=20
> assert can be triggered. If that doesn't give a solution in an hour or so=
,=20
> we'll just have to release with the problem still there. We could, just f=
or=20
> the release version, replace the assert by defensive programming, to avoi=
d=20
> the crash. I can't judge whether that's an option, because I'm not famili=
ar=20
> enough with the code.

It sounds to me like a race condition.  I don't know the code at all, but I
think putting a lock on m_commands might help.

Just in case I'm the only one who thinks about this (probably not ;-) ).

--=20
I encourage sending me encrypted e-mail.
Please send the central message of e-mails as plain text in the message bod=
y,
   not as HTML and definitely not as MS Word.
Please do not use the MS Word format for attachments either.
for more information, see http://129.125.47.90/e-mail.html