fix on SPARC+SolarisStudio

[iynehz]

Solaris Studio uses __sparc instead of __sparc__. And it does not have __arch64__.

Without this change it would fail t/checksum.t like #19

[karel-m]

Could you please post here the output of:

cc -xdumpmacros -E /dev/null

[iynehz]

#define __LINE__
#define __FILE__
#define __STDC__ 0
#define __STDC_VERSION__ 199409L
#define __DATE__ "May 19 2016"
#define __TIME__ "10:39:27"
#define __STDC_IEC_559__ 1
#define __STDC_IEC_559_COMPLEX__ 1
#define __STDC_HOSTED__ 1
#define __STDC_ANALYZABLE__ 1
#define __STDC_NO_ATOMICS__ 1
#define __STDC_NO_THREADS__ 1
#define __has_attribute (attr) ___has_attribute___ ( attr )
#define __SunOS_5_11 1
#define __SUNPRO_C 0x5130
#define __unix 1
#define __SVR4 1
#define __sun 1
#define __SunOS 1
#define __sparcv8 1
#define __sparc 1
#define __BUILTIN_VA_ARG_INCR 1
#define __C11FEATURES__ 1
#define __C99FEATURES__ 1
#define __PRAGMA_REDEFINE_EXTNAME 1
#define unix 1
#define sun 1
#define sparc 1
#define __RESTRICT 1
#define __FLT_EVAL_METHOD__ 0
#define __SUN_PREFETCH 1
#define __NOVECTORSIZE__ 1
# 1 "/dev/null"
#ident "acomp: Sun C 5.13 SunOS_sparc 2014/10/20"

[karel-m]

I have decided for a bit more complex redesign of tomcrypt_cfg.h

Could you please test CryptX-0.044_001 from CPAN (or the latest master from github).

[iynehz]

I tested your master from github, on SPARC+SolarisStudio. It's good.

[karel-m]

Great, I'll wait for cpan testers reports and check the new approach on couple of traditional trouble makers (AIX, HPUX & co.).

I will close this PR once I release v0.045

[karel-m]

Could you please check the latest devel version on your Solaris box

http://cpan.metacpan.org/authors/id/M/MI/MIK/CryptX-0.044_010.tar.gz

[karel-m]

fixed in CryptX-0.045