[Erp5-dev] Delayed reindex, workflow script and acquisition

[jp at nexedi.com]

Hi Bartek,

Please do not feel irritated. We were asked ourselves exactly the same
on zope-dev mailing list. It is quite normal.

Either your bug is the same as the one we fixed some months ago,
and it should be already fixed. This is why it is important that you
try recent releases of ERP5 (with official buildout). Or it is a different
one and we can not progress without a unit test from you. It may 
even be normal behaviour of MVCC is some cases.

Anyway, me or someone else we poste by tuesday the details of
what was fixed some months ago.

Regards,

JPS.



> On Sat, 13 Nov 2010 23:30:25 +0900
> Nicolas Dumazet <nicolas.dumazet at nexedi.com> wrote:
> 
> > On 11/13/2010 06:20 PM, Bartek Górny wrote:
> > > This goes back to my other email - the test to show the problem
> > > would be like:
> > > 
> > > doSomething
> > > doAnotherThing
> > > self.tic()
> > > check
> > 
> > Hello Bartek.
> > 
> > You seemed to care quite a lot about this issue in your previous
> > emails. From what I understand, you know enough of ERP5 to turns this
> > into a successful bug report.
> > I'm sure that you can afford to spend some more time on the diagnosis,
> > to prove us, with a complete TestCase, that something _is_ indeed
> > broken :)
> 
> Unfortunately, I can't. I've already spent enough time on
> it. What I'm writing about is an observation from a real-life
> installation. I know enough about ERP5 to tell you exactly what
> happens, when it happens, where it happens, to tell you how to
> reproduce the problem, and to formulate a hypothesis why it may happen.
> That's basically what a bug report is supposed to contain. What you are
> going to do with it is up to you.
> 
> See, I'm not lobbying for a feature request; I'm not asking you to do
> anything for me or for anybody. I'm warning you there might be a serious
> flaw in the core which in some circumstances can bring disaster. If
> I'm wrong, that'd be perfect - though I'd be more comfortable if
> somebody "from upstream" actually proved me wrong. If you don't care, or
> if you are so convinced that everything is perfect that you don't even
> need to check, good for you.
> 
> Bartek
> 
> > 
> > Regards,
> > 
> > > 
> > > to show what happens if "another thing" is done before the
> > > activities go through. I don't know any other way to reproduce this
> > > problem in tests.
> > > 
> > > Bartek
> > 
> > 
> 
> _______________________________________________
> Erp5-dev mailing list
> Erp5-dev at erp5.org
> http://mail.nexedi.com/mailman/listinfo/erp5-dev
>