Crypto++ 5.6.3 Release Candidate available (RC3)

[nolo...@gmail.com]

Hi Everyone,

I'm proud and embarrassed to say we have an RC3 available at http://www.cryptopp.com/cryptopp563rc3.zip. I'm proud because we've added a new platform. I'm embarrassed because it points to gaps in the engineering process, and the issue should have been detected sooner.

The new platform is a big endian, IBM S/390. László Böszörményi discovered the issue, and Michael Tokarev helped detect the S/390 environment. The issue was incomplete detection code in config.h because it appears the platform had never been tested or supported. The real iron is kind of rare, and the only place I'm aware it exists is Debian's s390/s390x porter machine (https://db.debian.org/machines.cgi?host=zelenka). Even GCC's test farm lacks it.

$ md5sum cryptopp563rc3.zip
02017d00bc74bb039d718d30613692

f8
$ sha1sum cryptopp563rc3.zip
5fc1cc71908b58d22cbed88d98ae224e371ed5ae
$ sha256sum cryptopp563rc3.zip
d715c07ce932726754816a993e4df7cd55c8e03799ab80e139ebf91084d308ea

Since the change was non-trivial, it kicked-off another round of testing. I've confirmed its compatible with other big endian platforms, like AIX and PowerMac; and does not interfere with little endian platforms, like Intel i686, x86_64, ARM, ARM64 and ARMEL.

Jeff