[Setup] Re: stack overflow (possible solved?)

[Alexander Limi]

Moving this over to plone-dev:

On Mon, 15 Jan 2007 08:16:01 -0800,  
<hpinson at indepthl.com> wrote:

> We may have found a solution to the Plone 2.5.2rc1 stack overflow
> problem.  I would like comments on our method and solution.
>
> We were able to reproduce the stack overflow error on a clean Plone
> site built in a pristine instance of Zope 296, Plone 2.5.2rc1,Python
> 2.4.4, and Fedora Core 6. No third party products were installed in
> this clean instance.
>
> The stack overflow was reproducable on some but not all pcs we
> tested. The common denominator seemed to be Windows XP, IE6 or 7, and
> maybe office 2003. I am unclear about what version of MSXML these
> systems were running.  We were never able to induce the stack
> overflow on any system, so the relevance of these variables is
> unclear.  The client either threw the error or did not.
>
> We installed a second pristine plone instance on BSD, with the
> similar configuration (using Plone 2.5.1 unfortunatly). This instance
> did not throw the stack overflow error on and client we tested.
>
> A diff revealed one javascript was being loaded on the Fedora
> instance that was not being loaded on the BSD instance.
>
> That script is named: event_registration.js
>
> Disabling that script in plone_javascripts seems to have solved the
> problem; no more stack overflow on clients that we can see.
>
> So my questions are:
>
> What does event_registration.js do?
>
> What have we broken by disabling it.
>
> Why would it load in a 2.5.2rc1 instance on Fedora Core 6 and not in
> 2.5.1 on BSD?
>
> What is it about the client systems that may or may not interact with
> this Javascript and throw a stack overflow error? Something Microsoft
> indtroduced into the mex very recently?
>
> Any thoughts are appreciated.
>
> I'm not sure if this is the right list for this discussion either...
>
>
>



-- 
Alexander Limi · http://limi.net