fix(cspell): sort and unique ignore lists

[OnkarRuikar]

The PR adds pre-commit hooks to sort and uniq the cSpell ignore list files.

[github-actions]

This pull request has merge conflicts that must be resolved before it can be merged.

[Josh-Cena]

Can you make this a CI check too? To avoid people bypassing it or committing from web interface.

[OnkarRuikar]

To avoid people bypassing it or committing from web interface.

Let's not prevent people from contributing. Like other stuff, we can always auto-fix it at midnight.

[Josh-Cena]

Most people do not change cspell files. Those that do (usually to address the weekly spellcheck errors) should ensure that they are changed in a correct manner.

Our Prettier review bot is the best contribution model. No bad content goes into the master branch and there's minimal friction to both contributors and reviewers. I think it should be implemented for everything, although Markdownlint is slightly different because the effect is non-local: a redirect can cause fixes in many files that you didn't touch. I would still advocate for local fixable changes (i.e. errors caused by changed lines) to be reported by the bot, though. And surely people change the dictionary file about once a week, so we should do everything properly in one commit.

[OnkarRuikar]

Most people do not change cspell files.

We mark the issues good first issues. So we should expect new contributors fixing typos.

How do you propose to handle this?

1. Sugeest sorted changes as reviewdog comments.
2. Create a new cspell list check workflow and fail it if files are not sorted properly and in logs ask the contributor to run the sort commands locally and push.

[Josh-Cena]

Unfortunately I would expect them to fix obvious typos but not, say, add things to dictionaries. It's too much of a judgment call to tell what to do with each word and I'm afraid, putting myself into the shoes of a new contributor, they would not have the courage to do. For example, who knows if I should add greetme to the dictionary or if I should change it to greetMe? Who knows if this abbreviation is actually a standard and/or intentional? Again I think we are too optimistic about how much freedom new contributors are willing to exercise—when I'm triaging issues, I do not put up "good first issue" unless new contributors know exactly which line to edit and what to change it to, without using any of their own discretion.

Ideally it would be reviewdog. But the changes are too non-local and result in proposed diff located in uncommentable areas. So, the workflow could output a whole dictionary file and ask them to copy&paste, or it could output a diff that we can apply (since ideally it would only be a few lines added each time anyway). What I don't want is two PRs every week, one called "fix typos" another called "fix dictionary file", both touching the same file. Do it once and do it right.

[OnkarRuikar]

Agree. Fixing typos in code is not good first worthy. I'll remove the good first label.

It's not possible to comment on unedited lines. I'll create a new check workflow that will run when the list files are edited.

[github-actions]

This pull request has merge conflicts that must be resolved before it can be merged.

[Josh-Cena]

Use Intl.Collator with sensitivity: "accent"

[OnkarRuikar]

The collator sounds good but cSpell doesn't seem use collator for comparisons. So after formatting the files using collator, I am getting the following errors from cspell because now the code treets 'acosh' = '𝚊𝚌𝚘𝚜𝚑'.

files/en-us/web/javascript/reference/global_objects/intl/displaynames/of/index.md:28:153 - Unknown word (nonterminal)
files/en-us/web/javascript/reference/global_objects/intl/displaynames/of/index.md:68:48 - Unknown word (canadien)
files/en-us/web/javascript/reference/global_objects/math/acosh/index.md:14:300 - Unknown word (𝚊𝚌𝚘𝚜𝚑)
files/en-us/web/javascript/reference/global_objects/math/asinh/index.md:14:204 - Unknown word (𝚊𝚜𝚒𝚗𝚑)
files/en-us/web/javascript/reference/global_objects/math/atanh/index.md:14:397 - Unknown word (𝚊𝚝𝚊𝚗𝚑)
files/en-us/web/javascript/reference/global_objects/math/cbrt/index.md:14:83 - Unknown word (𝚌𝚋𝚛𝚝)
files/en-us/web/javascript/reference/global_objects/math/expm1/index.md:14:83 - Unknown word (𝚎𝚡𝚙𝚖)
files/en-us/web/javascript/reference/global_objects/math/hypot/index.md:14:109 - Unknown word (𝚑𝚢𝚙𝚘𝚝)
files/en-us/web/javascript/reference/global_objects/math/log/index.md:14:182 - Unknown word (𝚕𝚘𝚐)
files/en-us/web/javascript/reference/global_objects/math/log10/index.md:14:182 - Unknown word (𝚕𝚘𝚐)
files/en-us/web/javascript/reference/global_objects/math/log1p/index.md:14:192 - Unknown word (𝚕𝚘𝚐)
files/en-us/web/javascript/reference/global_objects/math/log2/index.md:14:182 - Unknown word (𝚕𝚘𝚐)
files/en-us/web/javascript/reference/global_objects/math/sqrt/index.md:14:179 - Unknown word (𝚜𝚚𝚛𝚝)

The MathML pages use non-ASCII characters(a-zA-Z), that look like ASCII, in the formulas. I don't know the reason for using non ASCII abcd characters in the MathML code. Need to ask @fred-wang. Should we change those to make ASCII?

[Josh-Cena]

The MathML formulae often use the monospace Latin characters. Including JS pages I wrote, because I converted them using TeXZilla, so yes I'd be curious to hear what Frédéric says, including whether it's possible to make TeXZilla use inline CSS instead.

I'm surprised that 𝚌𝚋𝚛𝚝 and cbrt are considered equal though. Maybe we should keep using toLowerCase then—just change the regex to use a.toLowerCase() === b.toLowerCase().

[OnkarRuikar]

Now I've used toLowerCase for all the comparisons.

[Josh-Cena]

👍

[OnkarRuikar]

PR #36109 tested this PR check workflow.

• I created the Fix typos #36109 PR using the GitHub web interface.
• New words were added at the beginning of the ignore-list.txt.
• The workflow failed with the following error:

  The file is not formatted properly. Run 'node scripts/sort_and_unique_file_lines.js .vscode/ignore-list.txt' to format the file.
  Error: Process completed with exit code 1.

• Then I ran git fetch origin patch-3:patch-3 on my PC to get the PR branch.
• Then ran the suggested command to format the list file: node scripts/sort_and_unique_file_lines.js .vscode/ignore-list.txt
• Then pushed the changes using git push -u origin patch-3.