Consistent rules naming about typescript

[MartinJohns]

Adding even more options (or renaming them) will only bring more confusion, and you still need to consult the documentation to understand what the options do.

I've seen similar issues already, and they're always rejected. There gotta be a really good reason for such a change, and IMO you haven't presented any.

therefore obstructing intuitive and straightforward code styling.

It's not code, it's a configuration file.

[RyanCavanaugh]

Or maybe there is some specific reason why those rules were named like this?

They're named such that their default values are all false (or at least were at the time when they were named, though I don't think any have changed yet)

For example, if the default behavior is that foo is not allowed, then the flag is called allowFoo

If the default behavior is that bar is allowed, then the flag is called noBar

This keeps you from having to look up in the docs what the default value is - by looking at the name, you can tell.

Edit: The only flag that disobeys this rule is forceConsistentCasingInFileNames, which was false by default for a long time because there were performance problems associated with Node's syscalls for checking it. Once those were fixed, we changed it to true; its "proper" name would now be alllowInconsistentCasingInFileNames but this flag is so rarely set that it's really not worth introducing a second name for or introducing a config break.

[fatcerberus]

The self-encoding naming point is just a restatement of the bit you quoted.

[RyanCavanaugh]

I don't really see what the upside is here. There's already a completion list here; if you type "returns" you will see that it's noImplicitReturns not allowImplicitReturns.

Conversely, if two options that control the same thing exist:

• You have to figure out what it means if both exist
• You have to figure out what it means if both exist in different config files
• Basic string search no longer works
• Someone will log an issue saying you have to remove one of them "for consistency"
  • Then everyone argues about which one is correct
• Someone else logs an issue saying that all options should be phrased in a way that the default value of false is correctly interpeted "for consistency"
• Someone else logs an issue saying that they should not be renamed since the options in 5.5 were obviously good enough, and having two names is bad "for consistency"

[fatcerberus]

Let's just remove all the options and have tsc behave the same for everyone and be completely nonconfigurable... "for consistency" 🚎

[medreres]

@fatcerberus let's just be more friendly eh?

[medreres]

@RyanCavanaugh it's just I was studying ts config docs and it appeared to me counterintuitive to have such namings for rules in the first place. You made good point about this as well.

Someone else logs an issue saying that all options should be phrased in a way that the default value of false is correctly interpeted "for consistency"

Those are small things and are easily looked up in docs, but throughout time they gather like snowball and in the end you just have to consult docs for every such details and it's not a good developer experience IMHO.

Or maybe there is some specific reason why those rules were named like this? Or this is just a simple human error?

[fatcerberus]

let's just be more friendly eh?

@medreres That's the reason for the 🚎 emoji - trolleybus for friendly/playful trolling (and in this case I was trolling Ryan, not you). Don't take it too seriously

Those are small things and are easily looked up in docs, but throughout time they gather like snowball and in the end you just have to consult docs for every such details and it's not a good developer experience IMHO.

I would argue if you care about the options in a tsconfig at all you should be consulting the docs anyway, regardless of what they're named. The name alone can't convey all the effects a given option has, particularly in combination with other options, so what things are called is the least of your worries.

Or maybe there is some specific reason why those rules were named like this? Or this is just a simple human error?

More like "organic growth over time"; individual options are named pragmatically as they're introduced, naming preferences change, etc. etc. There's no specific reason for the haphazard naming beyond "that's just the way it worked out".

[RyanCavanaugh]

Note that this scheme is way better in terms of "saving you a trip to the docs" -- the completion list of options itself encodes their default values via the naming convention.

I think it'd be a lot worse if you were looking at a list of allowUmdGlobalAccess, allowPropertyAccessFromIndexSignature, allowImplicitOverride, allowUncheckedIndexedAccess, saying to yourself "Want that, don't want that, want that, don't want that" and having to cross-reference with the docs to see if you need to explicitly toggle them or not.

[medreres]

This keeps you from having to look up in the docs what the default value is - by looking at the name, you can tell.

Interesting thing, good to know, thanks :)

Didn't get the second point about self encoding naming though