For my purposes I'm not particularly satisfied. Though there's always a
possibility that it's operator error on my part. To name off a few
issues:
- I've yet to find a good way to force data to be consistently untyped.
Setting the "typed" configuration option doesn't seem to untype data
consistently, meaning I end up wrapping essentially every variable in
SOAPpy.untypedType().
- Why do I want everything untyped? Because SOAPpy's ints and longs
(which, as far as I know, just depend on type()) don't play nice with,
for example, Java/Axis's restrictions on ints and longs. Besides, often
with doc-lit services the type information is redundant.
- SOAPpy attempts to reference redundant values, possibly for
efficiency purposes. It does a poor job of this, often referencing the
wrong IDs, or attempting to link values recursively (i.e. dies with an
error). There's a configuration option to turn this off, but this
feature doesn't seem to be implemented. My workaround? Wrapping raw XML
in untypedType().
- It uses str() to marshal data.
I'd considered ZSI, but at the time it wasn't possible to set header
values, so it wasn't viable. It looks like they've updated since then,
so it may be worth re-evaluating. At present I use a heavily modified
copy of SOAPpy.
If anyone happens to have feedback regarding the above issues, I'd
appreciate the input.
Daniel O'Brien
m.banaouas wrote:
Hello Pythoneers:
I would like to ask you a simple question:
Are you satisfied by SOAPpy module?
Actually, I use this module since a year and i immediately found it very
well-implemented, corresponding to dynamic "meta-data" discovering
nature of SOAP/WSDL.
BUT i read sometimes threads saying that SOAPpy is "dead" and we should
all migrationg to Zolera SOAP Infrastructure (ZSI) module.
Can you tell me what do you think about it ?
thanks.
