Ed Jensen wrote: > Sion Arrowsmith <[EMAIL PROTECTED]> wrote: > >>I have here a library (it's the client side of a client-server >>interface including a pile of class definitions) which has >>implementations in pure C++, Java and Python, taking about 3000, >>3500 and 1500 loc respectively. And there's an associated module >>(with no C++ implementation) that I remember being particular >>"impressed" while writing it to find being 3 times as long in Java >>as Python despite (a) extensive (and pretty much common) Javadoc/ >>docstrings and (b) implementing in the Python version a feature >>present in the standard Java library (scheduling a thread to run >>at specified intervals and time out). Strip the Javadoc/docstrings >>out and it's about at that 5:1 ratio. > > > This claim seems pretty dubious to me.
i would not say sion's ratio of 5:1 is dubious. for what it's worth, i've written i pretty complex program in jython over the last year. jython compiles to java source code and the number of generated java lines to the jython lines is 4:1. bryan -- http://mail.python.org/mailman/listinfo/python-list