string.format() doesn't work under IE 11

[Egor-Skriptunoff]

Fengari is said to be "Verified to work in Microsoft IE 11".
But unfortunately an error is raised under IE 11:

print(("%02X"):format(0))
TypeError: Object doesn't support property or method 'repeat'

It seems that the sprintf.js which implements string.format is not IE11-compatible.
The problem is in the single line of code:

pad = ph.width ? (pad_length > 0 ? pad_character.repeat(pad_length) : '') : ''

Unfortunately, repeat is not implemented on all IE, including IE11

BTW .repeat occurs also in lstrlib.js.

Support of IE11 actually means support of Windows 7 users.
There are a lot of them nowadays.

All what is needed for happiness is to implement String.prototype.repeat manually in Fengari.

[daurnimator]

Note to self: see core-js implementation https://github.com/zloirock/core-js/blob/master/packages/core-js/internals/string-repeat.js#L12

[daurnimator]

Filed alexei/sprintf.js#182
Also see #149

[Egor-Skriptunoff]

I can't modify the global environment

But why?
Defining standard well-known (but undefined on IE) function looks very innocent.

[daurnimator]

I've worked in environments before where the rule is never modify the global JS environment. I want fengari to be allowed in such environments.

Every modification you make is potentially conflicting with some other piece of JS running on a site. The easiest solution is to just opt out.

[Egor-Skriptunoff]

Probably, this issue should be migrated from fengari to fengari-web, because it affects only fengari running inside browser.

[Egor-Skriptunoff]

The bug hasn't been solved in three months.
Maybe, the documentation should honestly claim that Fengari does require a polyfill for IE11?

[daurnimator]

I've been hoping that alexei/sprintf.js#183 will get merged.
Otherwise I may just remove that dependency entirely and implement our own sprintf-alike.

[daurnimator]

I haven't had the time to implement out own sprintf, so I'm marking this issue as "help-wanted". If anyone would like to implement it, please send a PR!

[catwell]

Would it be fine just to fork + vendor sprintf.js with your fix?

[daurnimator]

Would it be fine just to fork + vendor sprintf.js with your fix?

I guess. But there is plenty of code in there we don't use. And the verification code in lstrlib.js sort of duplicates some of the work anyway.
A rewrite with a bit of "inspiration" is probably the best choice.

[ashwalk33r]

I can't modify the global environment

But why?
Defining standard well-known (but undefined on IE) function looks very innocent.

You pollute users namespace. It's considered a bad practice. It's much easier to point users that it's not working on this and that browser and offer a link to a polyfill - if user wants then can include polyfil from separate repo - but then it's extra request or extra job to bake it into users code.

There is a way to create a copy of String prototype and assign it to new class - OurString and decorate it encapsulated only in module scope. @daurnimator what do you think?

---

or alexei/sprintf.js#193