LES: consider making `!!` a prefix/suffix operator like `++/--`

[qwertie]

LES: consider making `!!` a prefix/suffix operator like `++/--`

[jonathanvdc]

Oh, that's pretty interesting. Do you also want to integrate !!/!!. into EC#? If so, in what way would obj!!.field differ from obj.field?

[qwertie]

It probably doesn't make sense in EC# - in Kotlin there are separate data types Object and Object? so !! is needed to convert the latter into the former. EC# is not likely to add non-null types except to copy whatever C# does. I also consider LES as having higher priority since I need to make a bunch of decisions before porting the LES parser to other languages. I would welcome your input on the questions mentioned here... as well as the minus sign issue.

[jonathanvdc]

Your post about parsing negative numbers was an interesting read.

I think keeping LES as simple as possible is probably the best strategy. Always using string notation for negative numbers (i.e., number"-1" or i"-1", as you suggested in your post) seems like a solid strategy. My reasoning is this:

• People who want to use LES as a DSL (like you did in your graphing calculator) will probably end up implementing unary minus anyway, so they won't care whether LES parses -1 as @-(1) or as a -1 literal.
• People who want to use LES as a data storage format (this is actually pretty much what Flame IR uses LES for) won't care much either because their files will typically be computer-generated. They'll probably spend way more time reading LES files than writing them, and i"-1" is readable enough.

I'm not saying that parsing -1 as a literal has no valid use cases (manually writing a Flame IR file, for example, is one). But they really don't justify a complex implementation with funny edge cases in my opinion.

[Starwer]

Apparently, C# 8.0 will revolutionize the null-type by making all types value-types by default, unless declared otherwise using the ? postfix like in bool? myFlag;
They will also implement a ! meaning "I know this type is not null ! (throw exception at runtime otherwise)":
if (!string.IsNullOrEmpty(obj)) length = obj!.Length;

So I think this makes the proposed !! quite pointless.

BTW, even though I am not a user of this library, I'm positively impressed about the work done here.

[qwertie]

Hi Sebastien. Your comment would make sense in relation to Enhanced C#, but this is a thread about LES, which is conceptually unrelated to .NET. See http://loyc.net/les

[Starwer]

Indeed! My apologies ! I mixed up the two projects. Thanks for pointing out.

[qwertie]

!! and !. were actually added 2017-08-29 and I forgot to close the issue.