On 08/08/2016 20:31, Andy Seaborne wrote:
What do others on the PPMC think?
[...]
I remember we said the 3 most important points were
1. Community growth
Taverna is like many ASF projects - the size of the active (I)PMC is not
that great. This, in my experience, is normal. We hear and see the ASF
mega-projects but in terms of numbers of projects, they are a minority.
It would be good to see some of the IPMC active. The critical thing here
is whether people are available to get releases out, not "3 days, now"
but "in the next month could you...".
At the moment, my workload is pretty high with other things going on, so
I can only occasionally pay proper attention here. I'm afraid I've been
relying on others to pick things up and let me know explicitly when my
input is desired, and that's a bit naughty of me.
I'll parachute effort and attention in when I can.
2. Release more of the imported code to create engagement
What is the current state of imported/released?
There are two main items out of the imported set that haven't yet made
it to release: the Server and the Workbench. With the Server, I think
the effort to release it isn't too massive, under the assumption that we
don't take on doing a huge functionality revision. While there's some
bits that need work, I'm guessing that it isn't too much unless we go
for some of the more elaborate ideas that we've mooted in the past (and
I'm not convinced any more that they're the right way).
Concretely, the key things are:
* Review the internal message bus mechanism for security. JRMP is
convenient, but it requires very tight security and isn't really
designed to work that way. Attention required and not just from me
because I'm probably too close to the existing code to see any dumb
problems in it.
* Reworking the server so that it supports something less horrible
than baclava files for packaged data import and export. For export,
most people have been just downloading zip files, but the import
side is more of an issue.
* Throw out the mess that was the listeners and the notification
mechanism. That never really worked right. The bits that did work
are already mostly partially elsewhere, but we ought to clear this
bit of swamp instead of keeping the alligators for ass-backward
compatibility reasons.
Aside from the usual release engineering stuff (license checks, etc) of
course.
The Workbench is a whole different problem.
Donal.
