Hi all -
After months of threats, and months of head scratching, the Open Ernie site is now up and running at OpenErnie.org . Off we go to the races - I make the formal presentation at the VON Europe show tomorrow.
So, Ernie's mission is to enable LAMP developers to quickly develop VoIP applications using an architecture that scales. I've been looking at other open source efforts for raw material, including Asterisk, Yate, Shtoom and the offerings from SIP Foundry. Although the design stage will take some time, here's my thoughts on how things should be put together :
0) Ernie is open, free, standards compliant and quite SIPPY. Open means only open source will be used. Free means that no one should become the God of Ernie (that means me). Standards compliant and quite SIPPY means not only do we only implement with published and reviewed standards, but that we follow standard SIP and 3GPP architectures.
1) The SIP stacks from SIP Foundry are as good as it gets, commercial or open source. I can't think of a good reason to use anything else. There's too, too many SIP implementations already, so making a new one is just way out.
2) My personal feelings about Asterisk architecture are well documented, but the Asterisk developer community is large and vibrant. Many of them are taking Asterisk to make new applications. We need to find a way to integrate these efforts without inheriting the architectural problems.
3) Ernie should not be in any way a switch. It should be an applications platform, and that means that calls are present during the applications time, then should be ended or passed off. B2B = BAD
4) I think the basic top level API should be very close to TAPI in view and model, but implemented in Python/PHP/Perl. I happen to like Python very much. Es muy bueno.