davidiw
commented
12 years ago The reason why I used node wasn't out of preference, but because the session isn't defined until the desired group size has been reached. In the future, we may be able to construct a session earlier, but I wouldn't bank on it. We could listen to the signal from the node about being ready, but that would require extra work. I think this is one of those good enough things.
Have you experienced differently?
Hi David,
I spent this afternoon writing up some tests for the Web services. I tried to follow the test conventions where they exist, but I might have screwed it up. Can you take a look through here and let me know what you think?
Also: I changed SessionWebService services to use QSharedPointer -- not QSharedPointer, as you had made. My thinking is that it's better to give services the more restrictive access to session, and create NodeWebService for those ones that really need access to the whole node. Also, since the tests use TestNode, it makes it much easier to test Session services than Node services (since Node != TestNode). I can change it back if it's a big deal.
On the agenda for the next days/weeks:
Henry