[Zope-Coders] Re: test failures in TAL

[Martijn Faassen]

Steve Alexander wrote:
> Martijn Faassen wrote:
> >Jeremy Hylton wrote:
> >
> >>If I said the following, would it be sufficient:
> >>
> >>   These packages require older versions of Python, because they are
> >>   available as separate packages to customers running older versions
> >>   of Zope and/or Python.
> >
> >That would be true for *any* Zope product at all. Generally people use
> >older releases if they use an older Zope. This can't be the sole reason,
> >as it seems to be untrue for many other Zope packages. It also begs the
> >question as to when support for 1.5.2 will go away. Never?
> >
> >Anyway, I'd like to see this spelled out for each package seperately,
> >as it's obviously already raising 'why?' questions. 
> 
> I'd like to see some sort of exit condition on the support for Python 
> 1.5.2, such as
> 
>   "Foo Product will need to support Python 1.5.2 until all the Zope
>    Corp customers using Zope 2.3 have migrated to Zope 2.4 and above"
> 
> or whatever the reason is in a particular case.
> 
> Then, we have a testable condition to see when support for Python 1.5.2 
> can cease.
> 
> Even if the reason is "cos we say so", I'd like the requirement to 
> support Python 1.5.2 to be owned by someone:
> 
>    "Foo Product need to support Python 1.5.2 because Zope Corp says so.
>     The owner of this requirement is Joe Smith <j.smith@zope.com>".

This sounds good. In the case of "cos we say so" at least people
have the opportunity to bug the requirement owner. :)

It also has the consequence of making people think about the why of this
requirement a bit more, as they need to actually spell it out in some
shape or form.

Finally, there's a difference in support for Python 1.5.2 and support for
Zope 2.3 and less; one can write a product perfectly compliant with
Python 1.5.2 but requiring Zope 2.4 APIs. I think this also needs to be
spelled out clearly.
 
Regards,

Martijn