You made some reasonable points. I happen to disagree, but I admit that your points *could*
However, I'm interested in testing your hypothesis empirically. If you're right, then
Chandler would tend to have more runtime errors and take up more memory and CPU than a
comparable program written in C or C++.
Is there another such program that we could compare Chandler to? This is a serious question.
Hopefully our Grumpy Editor will check some out, and maybe when he does he'll pay attention to
the incidence of runtime errors and to memory and CPU usage.
By the way, I'm biased because I'm one of the authors of an app written in Python --
http://allmydata.org Tahoe, the Least-Authority Filesystem. We have shared some code and some
ideas with the Chandler project over the years.
If you find that Tahoe incurs runtime errors I would love to hear about it -- there are
currently no known errors of that type in Tahoe. This is probably due to our extensive test
suite which covers 93% of lines (http://allmydata.org/tahoe-figleaf/current/) and which gets
executed on eleven platforms on every code commit (http://allmydata.org/buildbot/waterfall).
Also if the memory or CPU usage of Tahoe is too great then I would be interested in that, too.
(We do use C or C++ for the CPU-intensive parts: cryptography, erasure coding, etc..) We have
automated measurements of the memory usage of Tahoe
(http://allmydata.org/tahoe-figleaf-graph/hanford.allmydat...) and I
would say it is currently "bad, but not critical", and probably no worse than a similar C++