El dom, 27-04-2008 a las 20:47 +0100, Ian Boston escribió: > So now we agree, the eco-system is more than just java. It would be > great to have multiple implementations of Shindig, as has started > with the PHP work, and where they need a build system they use > whatever is the most appropriate. Hopefully being an unbiased eco > system, some would evolve and grow and some will die. I guess all > depend on who wants to do the work, and who wants to use the outputs. > Surely that will drive what happens here ? >
+1, I just make noise as I don't have enough free energy to try a python implementation, which I would like seeing. > As a potential consumer of Shindig outputs, all I need is that it > runs well on Linux/BSD and talks over http. > Again +1, but see below for some long-term technical and non-functional concerns :) > When you say life outside java being fresher.... which language were > you thinking of? > Ian > I'm seeing a lot of activity using Ruby and/or Python, specially in the web front side. Modern frameworks such as Rails, Django, etc. In general diversity is increasing, but I've seen very interesting uses of those two, and I'm guessing (I wish I had hard data) than nowadays typical web sites get better sustained productivity if written with Rails,Django, etc. than with any of the "classic" java frameworks. In the back office, the main sources of innovation I'm seeing are erlang-based, things like CouchDB, with a very innovative architecture, or ejabberd, strong, scalable and lightweight. Also the use of python for infrastructural components, such as bzr or mercurial, is very interesting. Other people can bring more or better examples. But basically I'm more concerned about the recent conflicts re: the JSPA between the Apache Software Foundation and Sun Microsystem re: the JCK for harmony. Interpreting those as a struggle for control of java-the-language, my natural phobia for single failure points is triggered and I start looking for alternatives which are more distributed in nature. Of technical concern is the fact that the java VM (or C#, a java clone) microarchitecture has a very small granularity for state, and a very fine lock density (typically some mutable state and a lock per object). This plays very badly with modern hardware architectures, that are happier with mostly read-only, not locked memory and concentrated chunks of writable memory (erlang excels there, even if it has an ugly syntax and not so good libraries). Regards Santiago > > > On 27 Apr 2008, at 20:01, Santiago Gala wrote: > > > The reason why I'm bitching is not so much trying to get rid of maven > > here but try to open the eyes and minds of "world is java" developers > > to the fact that there is life outside java and that, increasingly as > > the java ecosystem freezes and ossifies, life outside java is fresher > > and evolving faster. No radical changes will work here, but to start > > moving such a big inertia requires a high amount of thrust and bashing > > ;-) And it seems like pushing in this direction is the only healthy > > behavior against lock-in, much like pushing for linux/BSD and against > > windows was the way to go in from 1995 onwards. For the health of the > > global computing environment. I might be wrong, of course. > > > > Regards > > Santiago > -- Santiago Gala http://memojo.com/~sgala/blog/
