Complex number in Scheme

[jcubic]

Information:

• Prism version: [e.g. 1.14, latest from the download page, etc.] Testing at website https://prismjs.com/test.html#language=scheme

Description
Complex number without imaginary number value is valid scheme complex number

Example
Example valid numbers that can be tested

10+i
10+.1i
10+1.i

[jcubic]

I did few attempts to fix the number regex, but it was failing for other tests.

[RunDevelopment]

Seems like I used an old version of the spec when making the number pattern. I'll fix it soon.

Note to me: Page 62 of this document defines the grammar for a valid number.

[RunDevelopment]

@jcubic I have a question. I implemented it and now 10i, 10.10i, and 10e+10i aren't tokenized as numbers anymore. This is because I implemented the BNF grammar on page 62-63 literally (you only need to look at the <complex R> rule). They aren't valid numbers according to R7RS because they don't start with a sign (either + or -). Should I continue to follow the spec or should I relax it a little, so that the above numbers are tokenized again?

[jcubic]

All examples works in Kawa but not in Gauche, Guile and Racket those are my base implementations that I'm testing against (racket is not exactly Scheme but it's compatible). The examples also work in my implementation (LIPS), except 10e+10i is parsed as 100.0, but that's a bug.

Maybe if Kawa support those, they can be marked as numbers, most of the time people will not use those type of values, because if they are not numbers they are treat as symbols so without quote '10 is marked as variable name, and they will use Kawa they would like to mark them as numbers.

[RunDevelopment]

If nobody is gonna these anyway, then I'll just keep the strict grammar.