[GHC] #10196: Regression regarding Unicode subscript characters in identifiers

[GHC]

#10196: Regression regarding Unicode subscript characters in identifiers
-------------------------------------+-------------------------------------
        Reporter:  thomie            |                   Owner:
            Type:  bug               |  thoughtpolice
        Priority:  normal            |                  Status:  patch
       Component:  Compiler          |               Milestone:  7.10.3
  (Parser)                           |                 Version:  7.10.1
      Resolution:                    |                Keywords:
Operating System:  Unknown/Multiple  |            Architecture:
 Type of failure:  GHC rejects       |  Unknown/Multiple
  valid program                      |               Test Case:
      Blocked By:                    |                Blocking:
 Related Tickets:  #5108             |  Differential Revisions:  Phab:D969
-------------------------------------+-------------------------------------

Comment (by thomie):

 Replying to [comment:5 hvr]:
 > We're planning to allow `Lm` from the 2nd character on in an identifier
 for 7.10.2

 The current patch does exactly this. It still needs a changelog entry.

 I would have preferred to only allow `Lm` in the suffix of an identifier.
 But we can leave that for 7.12 or later, as there is a slight chance it
 breaks someone's code. We could mention it in the docs.

 There's also the issue that ModifierLetter perhaps brings in too many
 weird characters:

 "15-06-18T11:46:27"<             hvr@> thomie: can we easily list all
 modifier letters in Haskell?
 "15-06-18T11:47:55"<             hvr@>  [ c | c <- ['\0'..],
 generalCategory c == ModifierLetter ]
 "15-06-18T11:47:56"<             hvr@> got it
 "15-06-18T11:48:56"<             hvr@> ok, there's a lot in there one
 doesn't want to allow in identifiers :-/
 "15-06-18T11:49:31"<          thomie > booh
 "15-06-18T11:49:50"<             hvr@> these look nasty:
 "15-06-18T11:50:46"<             hvr@> so many column variants, theres
 also "ː"

 hvr: do you think this a big enough issue to not proceed with the current
 patch?

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/10196#comment:9>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler