[GHC] #1473: isSpace is too slow

[GHC]

#1473: isSpace is too slow
-------------------------------------+-------------------------------------
              Reporter:  igloo       |            Owner:
                  Type:  bug         |           Status:  new
              Priority:  normal      |        Milestone:  ⊥
             Component:              |          Version:  6.6.1
  libraries/base                     |         Keywords:
            Resolution:              |     Architecture:  Unknown/Multiple
      Operating System:              |       Difficulty:  Unknown
  Unknown/Multiple                   |       Blocked By:
       Type of failure:              |  Related Tickets:
  None/Unknown                       |
             Test Case:              |
              Blocking:              |
Differential Revisions:              |
-------------------------------------+-------------------------------------
Changes (by dfeuer):

 * cc: hvr, ekmett (added)


Comment:

 I wrote up a whole long comment here, but then TRAC timed me out and I
 lost it. The gist of it is that I don't think we should be using
 `iswspace` here at all, because Haskell 2010's
 [https://www.haskell.org/onlinereport/haskell2010/haskellch16.html
 specification of `isSpace`] appears to be intended to match
 [http://www.unicode.org/Public/6.3.0/ucd/PropList.txt the Unicode concept
 of whitespace], which does not seem to match the locale-specific
 [http://pubs.opengroup.org/onlinepubs/009695399/functions/iswspace.html
 definition of `iswspace`]. As I discussed with Carter, we can probably
 improve the performance of `isSpace` by using something like Yitzchak
 Gale's implementation modified to cut down the number of branches (since
 many of them will be mispredicted).

--
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/1473#comment:7>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler