pep 8 constants

[Eric S. Johansson]

Peter Otten wrote:
> Eric S. Johansson wrote:
> 
>> MultiWordName  <cap>mulit<no-space><cap>word<nospace<cap>name
>> very high error rate.  many retries or hand hurting typing.
> 
> Can you define macros in your speech recognition software?
> 
> <cap>multi<camel>word<camel>name
> 
> might slightly lower the error rate.
> 

Yes it would. I think it would be possible to specify a better grammar however.
In the context of speech engine, if you know how the word is going to be used,
(i.e. it's a method, it's a class, etc.) you can automatically do the
transformation as part of the editors function. You need to know where you are
in the syntax tree and that gives you enough knowledge to do the name
transformation.

When you stop thinking of speech recognition interactions as discrete macros or
magic tricks, you can do a lot to accelerate coding.

Fruit equals pear tree sub branch plus 5

The translator should know that the name on the lval is a variable (type
signature determined later) and is terminated by the word equals (or =). The
system would then apply the appropriate the name transformation. Continue on,
equals would be transformed =, pear tree would be considered a complete name and
based on whether it is a class definition or instance, would be transformed as a
single name. Sub means there's an index here and would put the appropriate
brackets between the expression branch (symbol terminated by plus) and 5 (symbol
terminated by the end of line,

fruit = pear_tree[branch+5]

The next challenge comes in editing. It's fairly simple I would like to say
"edit line [1*<digits>]" and put that line in an isolated buffer where I can
edit the English form using all of the Select-and-Say controls.

That should close the cycle from creation through editing.  a small port of the
development cycle.

fyiw, the symbol trandformation code exists and has existed for almost 10 years.
 we need smart editing environments to make use of it.