Sheri schrieb:
> Christian Ehrlicher wrote:
>> Craig Silverstein schrieb:
>>
>>> } - check for stroll, strtoq, _strtoi64
>>>
>>> Do you check for "long long" as well?
>>>
>>> The tests will all pass if you don't do that check, but the system
>>> will be a bit crippled. Perhaps we should make it clearer when that
>>> happens (at build time?), so folks can look out for it. I expect
>>> "long long" to be pretty much everywhere. (Though maybe not on
>>> windows?)
>>>
>>>
>> the check for long long was already there and passed for msvc (don't
>> remember for mingw)
>>
>>
>> Christian
>>
>>
>
> Hi Christian,
>
> Some of the programs use such things as HAVE_LONG_LONG, which I don't
> see anything getting into
> config.h.
>
> Below are defines of probable consequence still not being done, except
> for the ones marked "now fixed")
>
> define HAVE_INTTYPES_H 1
> #define HAVE_LIMITS_H 1
> #define HAVE_MEMORY_H 1
> #define HAVE_STDINT_H 1
> #define HAVE_STDLIB_H 1
> #define HAVE_STRING 1
> #define HAVE_STRINGS_H 1
> #define HAVE_STRING_H 1
> (now fixed: #define HAVE_STRTOLL 1)
> (now fixed: #define HAVE_WINDOWS_H 1)
> #define HAVE_LONG_LONG 1
> #define HAVE_UNSIGNED_LONG_LONG 1
> /* #undef const */
> /* #undef size_t */
> /* #undef HAVE_READLINE_HISTORY_H */
> /* #undef HAVE_READLINE_READLINE_H */
> /* #undef HAVE_STRTOQ */
> /* #undef HAVE_ZLIB_H */
> /* #undef HAVE_BZLIB_H */
> /* #undef HAVE_DLFCN_H */
> /* #undef PCRE_EXP_DEFN */
> (now fixed: /* #undef HAVE__STRTOI64 */)
>
all except
#define HAVE_LONG_LONG 1
#define HAVE_UNSIGNED_LONG_LONG 1
are not needed by pcre.
I've attached a new config-cmake.h.in which add the two defines.
>
> Where are you heading on those prefixes? We need at least the option to
> make dlls in mingw and msys/mingw without "lib" prefix. All previous and
> current releases that included cmake build options did it for pcre and
> pcreposix without asking. (Problem was, it was also done for static
> libraries, and unix builds prior to latest). In my trial build with
> current revisions, it made libpcre.dll, libpcreposix.dll and libpcrecpp.dll.
>
Do you really need such an option? Why? It was wrong.
If you need it we can add an option explicit for mingw.
> This listng of options was nice, but I think one should say "-2" or
> preferably "ANYCRLF"
>
> Newline char/sequence ........... : 0x-2
>
fixed, the C compiler flags and C++ compiler flags are wrong (better:
not complete). Don't know if it's worth to spend time for this.
> I did see and use the option for rebuilding the character tables.
>
> Regards,
> Sheri
>
Christian
