Re: [Dev.team] Testing needed for Payforit

[Richard Frith-Macdonald]

On 21 Aug 2009, at 13:40, Polly Van Alstyne wrote:

> Olga and I did more tests today.  This time using the fake Payforit  
> channels Richard set up and also using the live Payforit channel
>
> Complete results are attached.
>
> On the live testing with a live channel, we had mixed results.  o2,  
> T-Mobile and Orange all succeeded, but vodafone and Three remained  
> unconfirmed.  Not sure if there's anything that needs to be looked  
> at here on those two.  Olga will try those 2 again later today to  
> see if the results stay the same.  Orange is an OBS transaction and  
> that one did go through, so it doesn't seem to be anything  
> specifically due to the OBS.

OK ... for vodafone we don't have an OBS process running, so we would  
expect all billing attempts to go nowhere.
For three we have a process running, but not connected to their  
billing system, so billing attempts would just keep retrying until  
they expire.

So for both of these, we would expect all billing attempts to be  
'unconfirmed' ... which is what you are seeing.  I don't think we have  
real connections to their billing systems do we?  Maybe channel 99  
(the one you are using for billing) should be configured with those  
two networks using fake billing connections?

Unless I've missed something, that leaves one major issue ... the  
constant refreshing on t-mobile.

I've changed the code in an attempt to fix this, working on the  
assumption that either the handset is buggy, or their proxy is buggy  
and keeps caching the page and returning the same cached page  
repeatedly.

The workaround is basically for us to use a new page each time rather  
than refreshing the same page (simply by appending a different  
sequence number to the path in the URL each time).  The hope is that,  
since each URL has a different path, no caching bug can break the  
refresh.
Unfortunately, doing that cleanly required quite a lot of changes to  
the code, so we need to retest extensively.  Are there any other bugs  
which need
quickly fixing before retesting ... we really need to make sure we  
don't change the code between final tests and the next release  
tomorrow evening, so if there are still things to do, I should get  
them done this morning before you do final tests.