XSD elements order must not be important about soapfish HOT 6 CLOSED

I think like option 1 as it seems this will require the last amount of changes for existing users and reordering could be useful for similar cases in the future. Option 2 doesn't sound too bad either, option 3 seems to be a very narrow solution (but might be worth it if the result code is ok to read).

Disclaimer: Please take these thoughts with a grain of salt. In the end we mostly need tests and maintainable code. I didn't dig too deep into the current code. If you feel that implementing nr 1 or so results in messy code, feel free to go after one of the other options. Also feel free to share "work in progress" code (e.g. by mail or in this ticket). I can't promise any timely feedback but at least I'll try.

from soapfish.

I didn't started it yet, but I think the option 1 is the best. Will try it this week and send some work in progress.

from soapfish.

It is done in my branch: FelixSchwarz/soapfish@master...iurisilvio:xsd_reordering

I don't like the solution yet. It works for my test, but I know some failing corner cases.

from soapfish.

I rewrote my implementation, using the default Python .sort(cmp). It looks better now.

I don't know every XSD schema detail, maybe it fails in specific cases (can I have a xsd:extension and xsd:restriction in the same complexType?), but I think it is good enough.

I'm trying it with my really complicated WSDLs to check if it fix my problem.

from soapfish.

Python 3 doesn't have the sort(cmp=) kwarg, but functools.cmp_to_key handle the problem nicely.

from soapfish.

I did the PR. It fixed my test and works with my WSDLs.

from soapfish.