concat to flags #17
concat to flags #17
Conversation
If we can do `RegExp.make("")("g")("i")("")…` anyway, let it be useful at least :-)
| if ('string' === typeof template && values.length === 0) { | ||
| // Allow RegExp.make(i)`...` to specify flags. | ||
| // This calling convention is disjoint with use as a template tag | ||
| // since the typeof a template record is 'object'. | ||
| return make.bind(null, template /* use as flags instead */); | ||
| return make.bind(null, flags + template /* use as flags instead */); |
There was a problem hiding this comment.
won't this cause a problem if i do make('g')('g')? or should we expect that to throw, much later, when it's invoked as a tagged template string?
There was a problem hiding this comment.
Ah, you're right, §21.2.3.2.2#7 says
if it contains the same code unit more than once, throw a
SyntaxErrorexception.
I didn't consider that. Do you think we should throw immediately? Filter out duplicates?
The change just looked simple and possible useful. I don't want to add any complexity…
There was a problem hiding this comment.
I'd prefer to throw immediately. If we know it will fail later, why defer the failure? The use case for currying here includes passing around the "maker" function for later use, so the exception could occur far away from the site of the double-flag.
Of course, another alternative is to uniquify the code units (perhaps [...new Set(flags + template)].join('')) in make.
If we can do
myMake = RegExp.make("")("g")("i")("")…anyway, let it be useful at least :-)I wonder what TC39 will say to this pattern, though. There are no curried functions in ES6 yet, and no variadic ones for sure.