Re: Check for redundancy

[Emanuel Berg <embe8573@student.uu.se>]

Stefan Nobis <snobis@gmx.de> writes:

> at least for some people it seems important enough

No doubt. I'm not in their exact positions (phew) so
I can't tell for sure but it seems I disagree
with them.

> Nevertheless, I think at least for junior
> programmers it might have some value (and for
> programmers in the industry with a high time
> pressure

For junior programmers it doesn't have a value save
for negative perhaps because junior programmers
shouldn't worry and feel disheartened by what they did
yesterday, they should just do new things, today!
If they later on feel like (manually) asserting old
code and make improvements and correct obvious
inefficiencies, that's on the other hand very helpful.

As for people in the industry working with high time
pressure whatever they do will never beat the thing
tons of people worked on, day in and day out, forever.
If they can remove redundancy from then next FPS game
with a movie licence I couldn't care less.

> its helpful to refactor redundancies with just
> a press of a button and adding some new names).

Indeed, the type of redundancy checker that will just
identify identical code blocks (yeah, two lines in
a row will do) - this is not a bad idea.

> Detecting patterns is one of the strong areas of
> machine learning and redundancy is a pattern.
> I would assume that AI methods will at least someday
> be quite good at it.

Is it really that difficult for a computer to do this?
The suggestion to have defuns byte-compiled
individually and then just compare the result sounds
much more straightforward.

Is there a prototype of that machine-learning
pattern-detector one can examine peace and quiet?

-- 
underground experts united
http://user.it.uu.se/~embe8573