Re: 2.1 changes - how to proceed (6 compile BUGS -with 1.3- in latest CVS)

[George I Matkovits <ma...@uswest.net>]

YES. I got it and it does NOT even compile. 6 errors. 3 in XMI Extensions (IMHO
Sanjiva made these) and 3 RETURN parameter casting errors in Template  and EJB
code modules.
The XMI ones are serious because they are in subclassing with a missing method
implementation. IMHO Sanjiva is the only one who 'really' knows what is going on
here.
What should we do? I think we should WAIT for the FIXES to be made ASAP and then
use this as our baseline. :-)
Regards - George
p.s. I am sending this to Doug directly, I will try Linux know with last
Friday's CVS.

Yaron Zakai wrote:

> George,
>
>  I understand that Wouter has committed all of his stuff to the CVS. I think
> that the best "baseline" we should use is the latest CVS, what do you say?
>
> Yaron.
>
> -----Original Message-----
> From: George I Matkovits [mailto:matkovitsg@uswest.net]
> Sent: Monday, January 08, 2001 4:38 PM
> To: soap-dev@xml.apache.org
> Subject: Re: 2.1 changes - how to proceed
>
> I used CVS from Saturday night + Wouter's 'better' fix. Without his fix the
> 'standard' CVS code dies with 2 addressbook requests. (Wouter's Saturday
> changes
> have not been committed yet!)
> I am going to run on Linux now (the only difference from your original code
> is
> that I use a DeploymentDescriptor.xml and deploy from the command line)
> Regards - George
>
> p.s. I will run 10*10000 loops on Linux RH6.2 and Java2-1.3 (JVM client
> optimization) with his 'better' fix (-: . I could not make it to fail with
> Wouter's 'better' fix, did you also use it? We should keep on using the same
> test for 'evaluation compatibility' reasons. You might also want to switch
> to
> the 'new latest' CVS + Wouter's fix.
>
> Yaron Zakai wrote:
>
> > George & Wouter,
> >
> >  During the last weekend I ran a bunch of massive tests, with the latest
> CVS
> > tree. I succeeded in reproducing the problem very rarely - it took 4-5
> > parallel sessions, and 100000 iterations to reproduce the problem. I'm
> sure
> > it exists, but it is very hard to reproduce, not to mention debugging.
> >
> >  What would you suggest to do for the next step?
> >
> >  Wouter - have you changed the way you are using the sockets in the last
> > round of changes which is not checked into the CVS?
> >
> > Regards, Yaron.
> >
> > -----Original Message-----
> > From: George I Matkovits [mailto:matkovitsg@uswest.net]
> > Sent: Monday, January 08, 2001 4:12 AM
> > To: soap-dev@xml.apache.org
> > Subject: Re: 2.1 changes - how to proceed
> >
> > Darn. (-: I was under the impression that you committed your changes on
> > Saturday night.) I got the CVS tree, added your 'better' fix and run
> > Yaron's  3 parallel * 10000 tests = 30000 test cycles OK. At least we know
> > that your 'connection' fix works perfectly :-) Could we get a parallel CVS
> > tree built somewhere?
> > Regards - George
> >
> > Wouter Cloetens wrote:
> >
> > > I've gotten very far this weekend. The changes are massive. I've
> > > touched over half the source files...
> > >
> > > I'm a bit reluctant to commit this stuff before it's properly tested.
> > > Especially, I'm not familiar with XMI, use of the envelope filter (I
> > > propose that as a new name for "transport hook extension" as it doesn't
> > > really work at the transport layer since the introduction of MIME
> > > functionality), the fault listener functionality, any of the scripting
> > > stuff and the SMTP transport. I don't have the setup to test EJBs
> > > anymore.
> > >
> > > Would you guys propose that I commit the changes, risk breaking some
> > > functionality or introducing bugs in the latest CVS, but fix those as
> > > people test and report problems, or should I make my in-progress tree
> > > available elsewhere and wait until it's stable to commit?
> > >
> > > bfn, Wouter
> > > --
> > > http://www.workspot.net/~zombie/soap/
> > > My opinions are irrelevant. They will be assimilated by my employer.