Re: [Bayonne-devel] problems with loopable script

[David Sugar]

The "pop" command removes one stack level.  If you have a ^xxx handler 
that catches an event, and then reruns the script or does a goto to 
another script, the stack is unchanged.  If you happen to be inside a 
for-loop, and a ^xxx handler like that is triggered, you have an extra 
stack frame that is not recovered when the script reloops or you branch. 
     If this happens repeatidly during the call, you will eventually 
use up the runtime stack (20 entries).  The pop command in the ^xxx 
handler can be used to manually get rid of the extra stack frame.  All 
this happens automatically in ccscript3.

Julien Chavanton wrote:
> Sorry I do not understand what you mean when you say 
> "use pop in your ^xxx handlers" can you explain as if this looks like a really interesting recommendation?
>
> The only 2 other problems left that I am aware of and I am going to patch on my 
> server so I can sleep well and start to look at Bayonne2:
>
> PROBLEM1 - Mysql/PostGres DB support without xml this sometimes cause crash 
> when Bayonne receive a hang-up interrupting an sql operation.
>
> PROBLEM2 - Dialogic digits capture using the lower level "dx_getdig()" instead of "dx_setevtmsk(chdev, DM_DIGITS)".  
>
> Going to Bayonne2 while leaving bug behind on my server in operation may be 
> lots of problems for me, since they will surely be work to do on Bayonne2 but I 
> am sure your new approach will be nice and efficient.
>
> Julien
>
>
>
> -----Original Message-----
> From: David Sugar [mailto:address@hidden 
> Sent: June 16, 2005 6:50 AM
> To: Julien Chavanton
> Cc: mailing_list_bayonne-devel
> Subject: Re: [Bayonne-devel] problems with loopable script
>
> I think if you have loops, and use pop in your ^xxx handlers, you should 
> be okay.  Automatic stack management changed bayonne scripting behavior 
> in other ways, and so I did not wish to bring it (or anything else that 
> changes existing application behavior) into the 1.x release series 
> unless it was absolutely necessary to do so.
>
> That being said, I think we need to review and come up with a new list 
> of open issues in 1.2 which still need to be addressed.  I am only 
> immediately aware of a few, but I would like a more comprehensive list. 
>   Other things which should be reviewed are questions like would it make 
> sense to backport the SIP driver to 1.2.
>
> Having regular releases for Bayonne 2 this early was done to avoid the 
> other mistake of the testing branch, by creating more pressure to 
> correct bugs and make it actually functional for use in deployed 
> configurations.  These things never happened for the testing branch 
> because it never was in a regular release cycle, so all the oddities 
> remained because they never actually "had" to be fixed.
>
> Julien Chavanton wrote:
>
> > Hi David,
> >
> > I did not have problem with loop yet, I must be really lucky, not sure why.
> >
> > Although you are planning to push Bayonne2 maybe we can try to correct the few 
> > last problems of Bayonne1.
> >
> > To support all the active installations and to be able to move to Bayonne2 
> > smoothly.
> >
> > Anything to say regarding this problem with loop?
> >
> > Julien
> >
> > -----Original Message-----
> > From: address@hidden [mailto:address@hidden On Behalf Of address@hidden
> > Sent: June 2, 2005 4:24 AM
> > To: address@hidden
> > Cc: address@hidden
> > Subject: Re: [Bayonne-devel] problems with loopable script
> >
> >
> >
> > Hi,
> >
> > David, how can we help you to debug it ?
> >
> > Do you need script examples with loop errors to test and trace ?
> >
> > Thanks,
> >
> > Emmanuel
> >
> >
> >
> >
> > On Thu, Jun 02, 2005 at 09:34:53AM +0200, Etoile Dièse wrote:
> >
> >
> > > Le Thu, 02 Jun 2005 09:15:59 +0200, <address@hidden> a écrit:
> > >
> > >
> > >
> > > > On Thu, Jun 02, 2005 at 10:15:22AM +0400, Dmitry Agafonov wrote:
> > > >
> > > >
> > > > > Hi!
> > > > >
> > > > > Bayonne (at least older releases) has a number of bugs in
> > > > > looping structures.
> > > >
> > > > Hi,
> > > >
> > > > I've got the same bug.
> > >
> > >
> > > Idem.
> > >
> > >
> > >
> > >
> > > > Is that bug in the 1.2.14 too ?
> > >
> > >
> > > I'm afraid, yes.
> > > I've got the bug in a section like :
> > >
> > > fordata #sql %var1 %var2 %var3
> > >         ...
> > > loop
> > >
> > > Sometimes (a few times per week), the ^error handler raises
> > > stack-overflow error
> > >
> > >
> > > F.
> > >
> > >
> > > --
> > > Etoile Dièse - www.etoilediese.com

dyfet.vcf
Description: Vcard