Re-allow trailing decimal points (e.g. `5.`) about json5 HOT 10 CLOSED

Uhm, why exactly are they disallowed? I think it makes sense to support them to denote that it's a floating number. node also seems to support them.

from json5.

@Midar, did you check out the discussion on pull request #9?

#9

JSON doesn't differentiate between ints and floats — it only defines a single "Number" type, and indeed, same with JavaScript — but is it the case that JSON implementations in other languages return specific types?

from json5.

Yes, other languages differentiate. In ObjFW, it's an OFNumber, but an OFNumber can contains a float, a double, an int, a size_t, etc. What it currently does is use an intmax_t for integers and a double for floating types.

Currently, it interprets 5.0 as a double and 5 as an integer. 5. would be a double then, while 5 would be an integer. It would actually make a difference if you go on and calculate something with it.

You can, of course, always convert it manually. The OFNumber class even has abstraction for that. For example, you call call [someNumber intMaxValue] and get an intmax_t - no matter whether it actually was an intmax_t or a double, it will just be converted if necessary.

However, these types have different limits, so it might actually make a difference.

What I also don't get about this is the inconsistency: Why would one allow .5, but not 5.?

from json5.

Good to know. I think you've made a reasonable case for allowing it. =)

from json5.

And here is another example:

@implementation Test
- (void)applicationDidFinishLaunching
{
        [of_stdout writeFormat: @"%@\n%@\n",
            [[@"[5.]" JSONValue] JSONRepresentation],
            [[@"[5]" JSONValue] JSONRepresentation]];
}
@end

Output:

[5.0]
[5]

I have to admit though that this is only since my last commit ;). OFNumber's description method was using the printf format specifier %f for floating points which would kill the decimal point if unnecessary. But since I fixed that bug in the description of an OFNumber, it behaves like pasted here. So I actually found and fixed a non-JSON5-related bug thanks to this discussion :).

from json5.

Happy to help! ;)

from json5.

Any progress on this? Can at least the spec be updated?

from json5.

Thanks for the nudge! Sorry there's been no progress on this; I've been quite busy. I can try to get to this over the holidays, but if you'd like to send a pull request, feel free! Even if it just updates the readme (but in that case, it'd nice to call out a TODO for this implementation to support it).

from json5.

Released in 0.2.0 FYI!

https://github.com/aseemk/json5/blob/develop/CHANGELOG.md#v020-code-diff

from json5.

Nice! Thanks!

from json5.