On Monday, March 24, 2014 11:37:05 AM UTC-5, Edward K. Ream wrote: >From another thread:
As some of you know, I recently restarted work on type-related programming > tools: > > https://groups.google.com/forum/#!topic/python-static-type-checking/Vhy_w5bCH8I > A brief update: I have a love/hate relationship with this project. Part of me wants to forget it entirely, the other part continues (annoyingly ;-) to be intrigued. To repeat what I've said before: tools like Rope and pylint already *are* static type checkers. I plan to add support to both pylint and Rope to Leo in the form of Leo commands. This should have a fairly high payoff. Both pylint and Rope follow same general strategy, namely "traversing/recursing" on the ast to attempt to infer type. Rope and pylint have already taken this strategy as far as it is likely to go. Both contain lots of clever code. I see little or no opportunity to do better. Imo, if there is any chance to create a new kind of tool, it must be based on rich global data. Preliminary measurements indicate that "big" data can be gathered a typical file in Leo's core in 0.1 seconds or less. I hope that this "big" data may be able to represent higher-level design criteria that are not easily detected in the ast itself. That's all for now. Edward P.S. Recent work convinces me that generators are an anti-pattern when creating global data. A symptom of the problems with generators is Pep 323: http://legacy.python.org/dev/peps/pep-0323/ For example, it turns out to be *impossible* to trace the intermediate results of a generator using Sherlock. Worse, generators make caching impossible. EKR -- You received this message because you are subscribed to the Google Groups "leo-editor" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/leo-editor. For more options, visit https://groups.google.com/d/optout.
