On 05/11/2012 03:19 AM, Kees Bos wrote:
On Fri, 2012-05-11 at 02:45 -0400, Stephen Waterbury wrote:
On 05/10/2012 11:18 PM, Lex Berezhny wrote:
On Wed, May 9, 2012 at 2:06 AM, Kees Bos<[email protected]> wrote:
Anyway, we've heard about companies thinking of dropping use of
pyjamas/pyjs, because of the uncertain future etc. This should be
addressed (in my opinion the future of the project depends on companies
using it) and embedding in a foundation would certainly help to gain
confidence in the future of this project.
Given that pyjs is currently a legal mine field I wouldn't be
surprised if some companies aren't using it for that reason alone.
Apache and GPL are incompatible licenses and yet there is quite a bit
of code in pyjs that's GPL.
Before there is talk of joining foundations, etc, it'd probably be a
good idea to actually make it safe and legal for companies to use it.
I think this discussion is somewhat of a red herring given there are
more basic issues that would prevent businesses from using pyjs. After
the legal problems are resolved the next top priority should be
quality. It's probably a safe estimate to say that half of the
messages on the mailing list in the last 6 months were people
complaining that a commit or merge to fix a problem/add a feature
broke some other feature or introduced a new bug. The developer rules
say that commits breaking functionality will be reverted but there
were at least a dozen times that even Luke had committed broken code
and nothing got reverted.
What we need is to hook up unit/integration tests to github pushes and
have a process in place that will actually do something about code
being committed that is broken.
I think there is a lot of work that still needs to be done before we
can confidently start shopping the project around to different
foundations for adoption.
Although personally I think that what pyjs needs a lot more than
adoption by big organization is for someone to write a killer app.
That will produce much better results since it will bring publicity
but also provide a good example of what a large application looks like
written in pyjs. I think that would be a better investment of time
than getting org adoption.
Amen to everything Lex says here. I've been using pyjamas/pyjs
for about a year and have developed a fairly complex in-house
application using it (over 4K lines of pyjamas code). As much as
I enjoy using it and think it has tremendous potential, I have
serious reservations about using it for new projects until I feel
more confident about its future. I think Lex put his finger
firmly on the top priority: quality management (unit tests,
perhaps even a buildbot or equivalent) should be first. Of
course a killer app could be huge, but many projects have been
quite successful without a killer app. However, an open source
project of the scale of pyjs can't succeed for very long if stuff
keeps breaking. Look at Twisted for an example of a strong
quality management system -- their level is very high and
probably overkill for pyjs at this point, but something to
emulate at least.
Steve
I think Lex is proposing quite some good stuff here, but I don't think
the time frame will be good enough for companies that currently cease
using it. I hope the PSF route will take about a week or so from now and
I know that this will help some people and companies.
Yes, having sponsorship of the PSF will help, but frankly, as happy as
I will be if it happens, I'm amazed that the PSF would agree to it.
Steve
