[PATCH] [2/6] Convert non-printing characters to their octal sequence before emitting #line directive or __FILE__ macro

[Yunzhong Gao]

  Hmm, something is messed up when sending out my email. I am trying again. Sorry.

  Replying to Arthur:
  > > This is part of a bigger effort to support foreign characters in file names.
  This sentence is merely to give a background of this patch, which is only the second of six patches...
  I am particularly interested in supporting Japanese shift-jis encoding (windows code page 932) on Windows. On these systems, #include directives will use UTF-8 encoding but file names on command prompt will use shift-jis encoding. Both will be translated to UTF-16/unicode before making system calls to the underlying file system.

  On a side note, I noticed that mails to
  reviews+D1291+public+fd336e303f55df78 at llvm-reviews.chandlerc.com
  do not appear in Phabricator as I expected. Maybe I should cc to cfe-commits.

  Thanks for reviewing!
  - Gao.


  > -----Original Message-----
  > From: Arthur O'Dwyer [mailto:arthur.j.odwyer at gmail.com]
  > Sent: Wednesday, September 11, 2013 11:37 AM
  > To: reviews+D1291+public+fd336e303f55df78 at llvm-reviews.chandlerc.com;
  > Gao, Yunzhong
  > Subject: Re: [PATCH] [2/6] Convert non-printing characters to their
  > octal sequence before emitting #line directive or __FILE__ macro
  >
  > > This is part of a bigger effort to support foreign characters in file names.
  >
  > You mean "to support filesystems whose character set is not Unicode", right?
  > I assume that on such systems, programmers would still write their
  > source code in UTF-8; it's just that they wouldn't have any control
  > over the encoding used for filenames in their filesystem.
  >
  > How does Clang currently deal with #include directives on such
  > systems? It seems to me that supporting #include should be much
  > higher-priority than supporting #line directives, and the same
  > mechanisms (UTF8 <==> filesystem
  > translation) could be used for both.
  >
  > What encoding in particular are you interested in?
  >
  > Alternatively, if this patch is merely about supporting "malformed
  > #line directives", you should say that instead. ;)
  >
  > –Arthur

http://llvm-reviews.chandlerc.com/D1291