Re: [PATCH] IBM z/OS + EBCDIC support

[Daniel Richard G.]

Hi Paul,

I tested your changes in git a406de9c. A handful of fixes are needed:

* c_isascii(): Add \x07 (DEL) as an ASCII character.

* c_isascii(): Drop \xFF (EO), as this is not ASCII.

* c_iscntrl(): Add \x07 (DEL) as a control character.

* c_iscntrl(): Drop \xFF (EO), as apparently this is not a control
  character.

* c_tolower(): In order to agree with tolower(), it needs to return the
  _unsigned_ promoted form of the character. (Returning identity in the
  default: case seems fine.)

* c_toupper(): Likewise.

* test-c-ctype.c: test_all(): As a result of the preceding two changes,
  this is needed:

  -          ASSERT (c_tolower (c) == 'a');
  +          ASSERT (to_char (c_tolower (c)) == 'a');

  -          ASSERT (c_toupper (c) == 'A');
  +          ASSERT (to_char (c_toupper (c)) == 'A');


Signed characters are a real PITA in EBCDIC. Even something like

    wchar_t buf[] = { 'a', 'b', 'c', '\0' };

doesn't work properly in that case.


--Daniel


On Wed, 2015 Sep 23 12:29-0700, Paul Eggert wrote:
> On 09/22/2015 11:58 PM, Daniel Richard G. wrote:
> >
> > What about having a check in test-c-ctype that compares c_iscntrl()
> > with its system counterpart? If the assumption is that alternate
> > EBCDIC encodings used with Gnulib will agree with EBCDIC-1047 on
> > these characters, then that should be checked.
> 
> Good idea.  Done in the attached patch.

-- 
Daniel Richard G. || address@hidden
My ASCII-art .sig got a bad case of Times New Roman.