2007/6/24, Jon Keating <[EMAIL PROTECTED]>:
Looks fine to me. So, where should we start coding? I have played around a bit
with the erijo-dev branch and it seems to be a decent start. If we start to
use that as a base, then I guess it would make sense to assign certain
portions to people. I will see about finally finishing the document about
owners, using the comments that I received awhile ago.

When newapi was first created we had a discussion if we should start
from scratch or use the existing code as a base. We agreed then that
the best way forward was to base it on existing code.

I then started erijo-dev and for some time I actually felt that
starting from scratch maybe wasn't that bad an idea. Then I reached
the point where the stuff I was working on was getting ready and I
wanted to test it on a bigger scale. That was when I realized that
working with existing code is that much better since you don't have to
write a lot of extra code just to test that one thing you're
interested in at the moment. So I think we should continue to use
newapi and I'll merge (part of) erijo-dev.

Maybe it takes longer time than starting from scratch (or not), but at
every step on the way, you can compile it, run it and get feedback
whether it works or not.

But I think we should use the structure from erijo-dev for new and
"cleaned up" code:
* The public api should consist mainly of interfaces and the header
files should go in licq/interface/<classname>.h
* Parts of the public api that are more of "helper"-class style
should have its header files in licq/<classname>.h
* Split logical parts of the source tree into src/<part> and have
separate unit tests for each part.

// Erik

--
Erik Johansson
http://ejohansson.se/

Reply via email to