Remove "sting" correction from dictionary_rare.txt

[TheGiraffe3]

"sting" is a word.

[DimitriPapadopoulos]

Indeed, sting is not even a rare word:
https://books.google.com/ngrams/graph?content=sting%2Cstring&year_start=1800&year_end=2022&corpus=en

I guess that this is a frequent typo in the context of source code, which was the initial target of this tool.

[TheGiraffe3]

Should I move it to dictionary_code?

[peternewman]

"sting" is a word.

Everything in the rare dictionary is a word, that's its literal definition!

codespell/codespell_lib/_codespell.py

Line 74 in e61ac16

"for rare (but valid) words that are likely to be errors",

You can either not use the rare dictionary, or exclude this particular word depending on your preference.

Indeed, sting is not even a rare word: https://books.google.com/ngrams/graph?content=sting%2Cstring&year_start=1800&year_end=2022&corpus=en

I guess that this is a frequent typo in the context of source code, which was the initial target of this tool.

I think the ngrams would look very different against a corpus of source code (if anyone has data for that).

As I raised ages ago ( #1469 ), should we split the rare dictionary. Tangentially also as @DimitriPapadopoulos raised, should it not auto-correct #3651 . I assume that happened and caused this PR @TheGiraffe3 ?

Should I move it to dictionary_code?

No, that's specifically for words that are valid in code, but not elsewhere (e.g. thead is a table header in HTML, but would be a typo elsewhere).

[TheGiraffe3]

No, I opened this PR because a GitHub Action is failing on one of my repositories every time I change a specific line (and forget to readd it to .codespell.exclude), and "sting" seemed to be a perfectly valid word to me.

I can close the PR (as with #1902) if you want, and I'll figure out how to disable the rare dictionary.

[peternewman]

No, I opened this PR because a GitHub Action is failing on one of my repositories every time I change a specific line (and forget to readd it to .codespell.exclude), and "sting" seemed to be a perfectly valid word to me.
I'll figure out how to disable the rare dictionary.

You can also ignore just the specific word sting if you'd prefer, which means you'll still get notified if you type chancel when you meant cancel for example, it depends on your appetite for potential false positives around the rare dictionary. Whereas (in theory) no valid words, but potentially some less frequently used domains-specific acronyms or abbreviations, should be in the main dictionary.