I get the same output with or without "g" flag at line 6:
https://run.dlang.io/is/9n7iz6

So I don't understand when I have to use "g" flag.

On Saturday, 23 June 2018 at 12:17:08 UTC, biocyberman wrote:
>
> I get the same output with or without "g" flag at line 6:
> https://run.dlang.io/is/9n7iz6
>
> So I don't understand when I have to use "g" flag.

My bet is that Regex results in D are lazy so "g" doesn't make sense in this context however I'm able to see an effect with "match":

match("12000 + 42100 = 54100", regex(r"(?<=\d)(?=(\d\d\d)+\b)", "")).writeln;
match("12000 + 42100 = 54100", regex(r"(?<=\d)(?=(\d\d\d)+\b)", "g")).writeln;

matchFirst would be like without "g"
matchAll would be like with "g"

On Saturday, 23 June 2018 at 13:45:32 UTC, Basile B. wrote:
> On Saturday, 23 June 2018 at 12:17:08 UTC, biocyberman wrote:
>>
>> I get the same output with or without "g" flag at line 6:
>> https://run.dlang.io/is/9n7iz6
>>
>> So I don't understand when I have to use "g" flag.
>
> My bet is that Regex results in D are lazy so "g" doesn't make sense in this context however I'm able to see an effect with "match":
>
> match("12000 + 42100 = 54100", regex(r"(?<=\d)(?=(\d\d\d)+\b)", "")).writeln;
> match("12000 + 42100 = 54100", regex(r"(?<=\d)(?=(\d\d\d)+\b)", "g")).writeln;
>
> matchFirst would be like without "g"
> matchAll would be like with "g"

I should have read the doc more thoroughly:

https://dlang.org/phobos/std_regex.html#match
>Delegating the kind of operation to "g" flag is soon to be phased out along with the ability to choose the exact matching scheme.

So case closed for me