incubator-wave-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ali Lown <>
Subject Re: Incubation status
Date Thu, 28 Nov 2013 14:18:48 GMT
>Playing the devils advocate I ask you (again):

Is this still Devil's advocate though? I have had a very similar email
sitting in my drafts for the last month asking the same questions
about the future of Wave.

>Do you folks believe the incubator can ever be completed as it is now?
>If you believe yes, please let me know why or how we can achieve that goal.
>Otherwise my recommendation is to move Wave to GitHub and close the incubation until the
community around Wave has grown.

I shall answer your questions throughout this email, though it
probably suffices to say that I no longer think Apache Incubator is
the right place for Wave (in its current form).
(With retirement: what happens to the project's source code license?
Does it become public domain instead of licensed to the ASF?)

>You already have it - wave on github. Here,

Yes, the code is on GitHub. (Though this is simply a one-mirror of the
Apache SVN tree).
[Though, if we retire the project that will no longer exist - I
suggest watching one of the personal trees (e.g. mine)].
When people are calling for GitHub, they are actually asking for the
development style that it uses: Git, Pull Requests, Quick-forking,
Less 'paperwork'. [And to some extent the 'coolness' factor - which is
not to be underestimated for getting development support].

>lets finally have discussion for development happen on a public wave ;)

I agree that the dogfooding should really have been a thing, but it
hasn't been possible here. (Though I hestitate to say whether Wave is
stable enough for multiple users heavily editing a Wave - my anecdotal
data says it tends to 'get stuck' around the 100 blips mark).

> Speaking as someone unable to contribute code to the client as its too
> heavily tide into the server (which I cant make heads not tails of),

This is a major contention point. It is definitely too tied together,
but because of this, it is very difficult to separate it now... (But
this is something that must be done).

>how will any move effect things? how will it help? wont it just be rearranging
> things again that have little, if anything, to do with getting anything
> actually done?

It would indeed seem mostly arbitrary with regards to the tooling. The
ethic however is quite different for GH projects, compared to Apache
projects. (And I would argue it is this, that is part of the reason we
struggle to maintain active developers here).

The other problem, is that at ~500,000 LOC of Java, it is not easy for
new people to get involved. (@Ewan: This ties in to your point, but it
would take more than a few weeks to get someone familiar with this
codebase [I have been focused almost exclusively on the server code
for the last ~3 years, but I still couldn't tell you exactly how it
all fits together - which is why the corruption issues are still

> I am still massively enthusiastic about WFP as a communication method, and
> making a good reference client and server is the way to push it.

This I agree with, but it also tells us what our actual aim should be:
A clearly separated library for using WFP to create things - of which
the client/server are examples...

Ultimately, from my point of view, a move to GitHub would provide us
with several things:
- Full Git integration (The Apache system is still very awkward to use
and git-svn still chokes on things occasionally).
- The GitHub 'ethic' - hard to explain
- The opportunity to change the working style. I feel that the
'meritocracy' approach only works well for clearly established
projects. Wave has too many options - and it is this that is dividing
the effort going in to it. Making decisions here is proving incredibly
difficult, getting votes for releases is very difficult, etc. As such,
I would push for a much clearer philosophy of the 'new project'.

Sorry about the long email. :)


View raw message