I attended the SDForum Dynamic Language Series on Python this morning. I'll jot down a few of the highlights here.

Guido and the history of python

The program started off with a history of python from Guido. He had two weeks of free time and wanted to implement a language. He said his main design goals were interactiveness, work well with other tools, avoid NIH (borrow freely), and doable as a one person job. The last one was a little interesting since he explicitly said he would have to cut corners to make it a one person job (seems like this is common amoung new open source projects since the normally have only one dev). He also said that the object oriented nature wasn't due to a deep theoretical understanding of object oriented principles but that it grew out of a need to enable extensibility in the language.

Python in the Enterprise

This was a panel discussion called "Python in the Enterprise" by Munwar Shariff (CTO Cignex, a Zope/Plone shop) and Alex Martelli. The moderator (who was an editor for Software Development, a programming magazine) said she noted that python programmers net about $10K/year more than most programmers. Alex attributed that to being more productive with python.

There was a statement that python's biggest problem was that it didn't have a compiler and hence the source was available. Alex said people can decompile anything, so the ONLY way around this is to provide a web service if you really don't want your IP in the wild. Munwar stated that he had sold product to many government agencies (but his product is also a web service).

There was the standard lack of static type checking complaint. The reply was Eckels quote that code that isn't tested is worthless anyway (I'm paraphrasing a little).

I asked what the panelists thought when I recently asked a friend what was the difference between two products (one in java and one in python), and he (he happens to sell the java version) replied "It's written in python, enough said", implying that no one would buy the python version by virtue of the language it was written in. I don't think Alex understood what I was trying to get at, which was the ingrained notion that enterprise software has to be written in java (perhaps it's analagous to "no one ever got fired for buying from IBM", choosing Java software is a safe choice because you don't have to justify it). Alex said, "Don't tell them what language it is written in." He also mentioned that you could use jython too. Munwar admitted that they had lost a sale because they were pushing the zope platform.

There was talk was actual research that Python was "better", rather than anecdotal evidence. Alex mentioned something report by Franheufer Institute? (I'll have to see if I can locate it). Location is here Munwar offered his anecdotal evidence that developers were 5-10x faster in python, but that performance in python suffered a tiny bit. Stay tuned, I'll try to post a followup with Greg Stein's "Python at Google" and some notes from the "Python in startups" panel soon.