[Python-Dev] How far to go with user-friendliness

Steven D'Aprano steve at pearwood.info
Fri Jul 17 18:20:52 CEST 2015


On Fri, Jul 17, 2015 at 04:37:04PM +1000, Nick Coghlan wrote:

> The specific typo that is checked is the only one that changes the
> spelling without also changing the overall length and shape of the
> word.

I don't think your comment above is correct.

    assert => aasert aseert azzert essert assort

all have the same overall length and shape.

Not all spelling errors are typos (hitting the wrong key). I've seen 
spelling errors this bad, or worse, from native English writers. Poor 
spelling, bad keyboards, distraction, and dyslexia can all contribute. 
And those who aren't fluent in English will make their own spelling 
errors, and may not even notice if the length of the word changes:

    assert => asert

For those who are dyslexic, there are spelling errors and typos that may 
be difficult to tell apart even though the shape of the word changes:

    assert => assery asserh 

(perhaps -- I'm not dyslexic, I'm just going by what I've read about 
their experience).

In my opinion, this sets a bad precedent for adding special case after 
special case, and the risk is that people will feel slighted if they are 
told that their typos aren't important enough to be made a special case 
too.

If Michael wishes to argue that this is a useful feature rather than an 
ugly DWIM wart, that's his perogative, but the justification that 
"assret" is the *only* such plausible typo is just plain wrong. We've 
already heard from Robert Collins that he found a bunch of silently 
failing assertions in his mocks, and none of them started with "assret".


All-spelling-errors-are-deliberate-to-provide-new-and-exciting-ways-to-spell-old-words-ly y'rs,

-- 
Steve


More information about the Python-Dev mailing list