Date: Tue, 3 Sep 2002 17:11:10 +1000 (EST) From: Bruce Evans <bde@zeta.org.au> To: Mike Barcroft <mike@FreeBSD.org> Cc: standards@FreeBSD.org Subject: Re: Conformance bugs in <ctype.h>, by way of <runetype.h> Message-ID: <20020903165059.Q6101-100000@gamplex.bde.org> In-Reply-To: <20020902140549.E94253@espresso.q9media.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, 2 Sep 2002, Mike Barcroft wrote:
> >From rev 1.5 of ctype.h:
> ...
> o TODO: fix conformance bugs brought by including <runetype.h>.
>
> I'd appreciate comments of the attached patch which fixes this issue
> and brings <ctype.h> up to conformance.
> ctype.diff
>
> o Fix namespace scope issues in <ctype.h> by using the relatively new
> visibility primitives.
> o Implement _tolower() and _toupper() POSIX.1-2001 (XSI) macros in
> <ctype.h>.
> o Make <runetype.h> pollutant-free so that it can safely be included
> in <ctype.h>.
> o Add a typedef in <rune.h> to compensate for <runetype.h> being
> pollutant-free.
>
> Index: ctype.h
> ===================================================================
> RCS file: /work/repo/src/include/ctype.h,v
> retrieving revision 1.22
> diff -u -r1.22 ctype.h
> --- ctype.h 21 Aug 2002 16:19:55 -0000 1.22
> +++ ctype.h 2 Sep 2002 18:07:04 -0000
> ...
> @@ -85,9 +81,15 @@
> int tolower(int);
> int toupper(int);
>
> -#if !defined(_ANSI_SOURCE) && !defined(_POSIX_SOURCE)
> -int digittoint(int);
> +#if __XSI_VISIBLE
> +int _tolower(int);
> +int _toupper(int);
> int isascii(int);
> +int toascii(int);
> +#endif
It would be nice to not support mistakes like XSI.
> Index: runetype.h
> ===================================================================
> RCS file: /work/repo/src/include/runetype.h,v
> retrieving revision 1.7
> diff -u -r1.7 runetype.h
> --- runetype.h 21 Aug 2002 16:19:55 -0000 1.7
> +++ runetype.h 2 Sep 2002 17:18:18 -0000
> ...
> @@ -67,9 +50,9 @@
> * The lower 8 bits of runetype[] contain the digit value of the rune.
> */
> typedef struct {
> - rune_t min; /* First rune of the range */
> - rune_t max; /* Last rune (inclusive) of the range */
> - rune_t map; /* What first maps to in maps */
> + __rune_t min; /* First rune of the range */
> + __rune_t max; /* Last rune (inclusive) of the range */
> + __rune_t map; /* What first maps to in maps */
> unsigned long *types; /* Array of types in range */
> } _RuneEntry;
>
> ...
There is still a lot of pollution here: all the struct member names, which
don't even have a runetype-specific prefix (#define min runetype loses).
There doesn't seem to be any good way to fix this. The inline functions in
ctype need to reference at least the runetype[] struct member. Renaming
the struct members to have a prefix like __ru_ would be ugly and might
break documented runetype interfaces. Most of the interfaces seem to be
private so this might not be a problem.
Bruce
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-standards" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020903165059.Q6101-100000>