Thread overview
Newline character set in the D lexer - NEL
August 31
Would there be any benefit from the following suggestion? Add the character Unicode NEL U+0085 into the set of EndOfLine characters in the lexer ?

Cecil Ward.
August 31
On Monday, 31 August 2020 at 01:49:06 UTC, Cecil Ward wrote:
> Would there be any benefit from the following suggestion? Add the character Unicode NEL U+0085 into the set of EndOfLine characters in the lexer ?
>
> Cecil Ward.

I personally think we should have these definitions:

             /*  NUL    EM    SUB */
EndOfFile   = { 0x00 | 0x19 | 0x1A | PhysicalEndOfFile };
             /*  LF     FF     CR      CR LF     NEL     LSEP     PSEP  */
EndOfLine   = { 0x0A | 0x0C | 0x0D | 0x0D 0x0A | 0x85 | 0x2028 | 0x2029 | EndOfFile };

             /*  HT     VT     SP    NBSP    NQSP     MQSP     ENSP     EMSP     3/MSP */
WhiteSpace  = { 0x09 | 0x0B | 0x20 | 0xA0 | 0x2000 | 0x2001 | 0x2002 | 0x2003 | 0x2004

             /*  4/MSP    6/MSP     FSP      PSP     THSP      HSP     ZWSP     NNBSP */
              | 0x2005 | 0x2006 | 0x2007 | 0x2008 | 0x2009 | 0x200A | 0x200B | 0x202F

             /*  MMSP      WJ      IDSP    ZWNBSP */
              | 0x205F | 0x2060 | 0x3000 | 0xFEFF | EndOfLine };

The definition of D source files misses quite a lot of them :-(

EM = end of medium (what if not this should end a file?!?)
NEL = New Line
LSEP = Line Separator
PSEP = Paragraph Separator

NBSP = non-braking space
NQSP = ENSP = N-wide space
MQSP = EMSP = M-wide space
3/MSP = 1/3 M-wide space (three spaces together are as wide as an M)
4/MSP = 1/4 M-wide space
6/MSP = 1/6 M-wide space
FSP = figure space
PSP = point space
THSP = thin space
HSP = hair space
ZWSP = zero width space
NNBSP = narrow non-braking space
MMSP = mathematic space
WJ = word joiner (invisible space that separate words for the spelling correction)
IDSP = ideographic space (same width as a chinese character)
ZWNBSP = zero-width non-braking space
August 31
On Monday, 31 August 2020 at 01:49:06 UTC, Cecil Ward wrote:
> Would there be any benefit from the following suggestion? Add the character Unicode NEL U+0085 into the set of EndOfLine characters in the lexer ?
>
> Cecil Ward.

Pardon me but why bother while ascii gives already all we need to put spaces and new lines with fast decode (< 80h) ?
August 31
On Monday, 31 August 2020 at 09:39:12 UTC, Nils Lankila wrote:
> On Monday, 31 August 2020 at 01:49:06 UTC, Cecil Ward wrote:
>> Would there be any benefit from the following suggestion? Add the character Unicode NEL U+0085 into the set of EndOfLine characters in the lexer ?
>>
>> Cecil Ward.
>
> Pardon me but why bother while ascii gives already all we need to put spaces and new lines with fast decode (< 80h) ?

D already recognizes some non-ascii characters as spaces and line separators [1], so the decision to "bother" has already been made.

[1] https://dlang.org/spec/lex.html#character_set
September 04
I agree with Dominikus

Note to earlier poster: NEL was used and just possibly may still be used by IBM mainframe users; XML 1.1 understands NEL iirc;

see https://www.w3.org/TR/newline/   and

       https://www.w3.org/International/questions/qa-controls
September 04
On Friday, 4 September 2020 at 00:48:59 UTC, Cecil Ward wrote:
> I agree with Dominikus
>
> Note to earlier poster: NEL was used and just possibly may still be used by IBM mainframe users; XML 1.1 understands NEL iirc;
>
> see https://www.w3.org/TR/newline/   and
>
>        https://www.w3.org/International/questions/qa-controls

Given the lack of answers I would suggest to go ahead with a PR or at least open an issue. Lexing is not a big deal but if nobody cares this will never be done.
September 08
On Friday, 4 September 2020 at 05:28:47 UTC, NilsLankila wrote:
>
> Given the lack of answers I would suggest to go ahead with a PR or at least open an issue. Lexing is not a big deal but if nobody cares this will never be done.

Agreed, Nils. Mind you someone cared enough to include U+2028 and U+2029 in the lexer spec.

I have no idea how to initiate a "PR". Perhaps someone could help me with this?
September 08
PR = "Pull Request".

Easy way is to fork the project on github, clone your (forked version of the) project, make changes, push back. This could be in ~master on your own fork, or ideally in a separate branch.

Then on github, go to the original project and start a new pull request. It should automagically detect that you've made changes (again ideally in a branch of your fork), and offer to make a pull request with your changes against ~master (or whatever is set as the default branch for the project).

James

On 9/8/20 2:42 AM, Cecil Ward wrote:
> On Friday, 4 September 2020 at 05:28:47 UTC, NilsLankila wrote:
>>
>> Given the lack of answers I would suggest to go ahead with a PR or at least open an issue. Lexing is not a big deal but if nobody cares this will never be done.
> 
> Agreed, Nils. Mind you someone cared enough to include U+2028 and U+2029 in the lexer spec.
> 
> I have no idea how to initiate a "PR". Perhaps someone could help me with this?