[XML-SIG] yet another output encoding

[Walter Dörwald]

M.-A. Lemburg wrote:

 > [...]

> 
> I thought about such a possibility today too: maybe all
> we need is a simple registry of error handlers. The handlers
> are then addressed using their string name and we can leave
> the existing APIs (which use const char *error parameters)
> in place.
> 
> Is that what you have in mind too ?


Exactly.

The encoder would still have to (re)allocate the unicode
object passed in as Py_UNICODE*/int but only once and only
for unknown error handling names.


>>Now the only remaining problem is time! :-/
>>
> 
> Python 2.3 will leave us around 6 months -- should be 
> enough :-)


I hope.

Bye,
    Walter Dörwald