Language change and code breaks (fwd)

[Bjorn Pettersen]

> From: Nick Efford [mailto:nde at comp.leeds.ac.uk]
> 
> On Tue, 17 Jul 2001 20:09:05 -0400,
> Lulu of the Lotus-Eaters <mertz at gnosis.cx> wrote:
> 
> > I--one of those professional programmers--write a Python script for
> > someone who runs a business, or a non-profit, or an agency, 
> to run on
> > their CGI webserver.  This client knows not the first thing 
> about Python
> > or about programming, nor do they maintain the webserver, just rent
> > space on it.  But still, this script does something useful 
> for users of
> > the website--order products, or request documents, or send data to a
> > database.
> > 
> > I go away, no longer available.  The client's web hosting company
> > upgrades the Python installation to Python 3.7 (this is in 
> the future
> > :-)).  Client's scripts break.
> 
> Interesting scenario.
> 
> Of course, this could happen with other s/w, too.  This really
> says more about the policy of the web hosting company w.r.t.
> upgrades than it does about the wisdom or otherwise of breaking
> compatibility in a future version of Python.
> 
> Any web hoster worth its salt should at least run old and new
> versions of the s/w in parallel and provide clients with the
> opportunity to test programs in the new environment before
> migrating.

But being a web hoster, with no knowledge about Python, would _you_
expect a version change from 2.2 to 2.3 to fundamentally change
semantics? If you're really unlucky with this proposed change, your
program could continue to run but silently give you wrong answers. (I
can understand being extra careful with going from e.g. 2.2 to 3.0
however...)

-- bjorn