[Soap-Python] SOAP Fault

[Burak Arslan]

On 04/10/12 16:56, Stefan Andersson wrote:
> On 4 October 2012 17:27, Burak Arslan <burak at arskom.com.tr> wrote:
>> On 04/10/12 14:13, Stefan Andersson wrote:
>>> Hi,
>>>
>>> I have fault type now with support for other subelements both in
>>> respons and wsdl, but now when I synced up with your master repo I get
>>> circular import problem because of changes in model/complex.py, my new
>>> fault.py imports ComplexModelMeta and now complex.py imports stuff
>>> from primitive and primitive imports fault.
>>>
>>> The way I see around this is to implement a copy of ComplexModelMeta
>>> in fault or try break it out from complex.py without any dependencies
>>> on complex.py, what would you prefer?
>>
>> Can you show me your work so far? Why did you need ComplexModel machinery in
>> Fault?
> I have commited it to the branch restriction (commit:
> https://github.com/norox/spyne/commit/6878512695aa3db1ffbcfdc12c90c740072bfdce)


Hi,

First, why not base on ComplexModelBase if your metaclass is 
ComplexModelBaseMeta?

Common sense says that _base -> primitive -> complex -> fault is a 
plausible dependency graph and the thing that is actually out of place 
here is primitive importing fault.

So "the right thing" to do here is to move the {to,from}_string logic to 
ProtocolBase to avoid primitive importing fault.

If that'd be too much work for you,

_base -> primitive -> complex
      \-> fault

also sort of makes sense (this is currently the case without your 
patch), but you'd have to re-implement a ComplexModelMetaLite as moving 
some of the code in complex.py to another module won't break the circle.

This is a hack and I would not really prefer this solution. I also don't 
think it'd be less time-consuming because the spyne.mode.complex module 
is quite mature as it covers a lot of corner cases that took a while to 
run into.

So all things considered, I think you should follow the first option.

Best,
Burak

>
>>
>>> //Stefan
>>>
>>> On 2 October 2012 12:56, Stefan Andersson <norox81 at gmail.com> wrote:
>>>> Ok, I will take look and see how much work it is, would make
>>>> integration with some system we have easier.
>>>>
>>>> On 2 October 2012 10:54, Burak Arslan <burak at arskom.com.tr> wrote:
>>>>> On 02/10/12 09:48, Stefan Andersson wrote:
>>>>>> Hi,
>>>>>>
>>>>>> Do you know if anyone have been looking into adding other subelements
>>>>>> to SOAP Faults?
>>>>>> According to
>>>>>> http://www.w3.org/TR/2000/NOTE-SOAP-20000508/#_Toc478383507
>>>>>> it's supported, "Other Fault subelements MAY be present, provided they
>>>>>> are namespace-qualified" (in the end of 4.4 right before 4.4.1).
>>>>>
>>>>> Hello,
>>>>>
>>>>> No, and I had no idea it was allowed (must have missed that line).
>>>>>
>>>>> Burak
>>