-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

This whole thing to do with crashes in release builds of crypto++ 5.0 is 
actually rather worrisome.

If one releases code without the NDEBUG flag, then, because not all asserts 
were replaced by exceptions, then there is a non-zero chance that the code 
will hit an assertion failure [because bad/inconsistent data might be 
passed to a crypto++ routine] and simply abort. (I had this happen to a 
customer this weekend.) But if one uses the NDEBUG flag, then the code may 
SEGV.

Am I alone in worrying about this? 

Are there any plans to remove all the remaining assertions and replace them 
with thrown exceptions?

  Doc


-----BEGIN PGP SIGNATURE-----
Version: PGP 6.0.2 -- QDPGP 2.60 
Comment: Key obtainable from servers: ID 0x362912B8

iQA/AwUBPeuCnGnXrLw2KRK4EQIi6gCeMpW83dm9crjvd2wy5SVcFSUh0KgAoI9Y
K7lv308HkHp8wjcvuaUWfpgh
=pxt9
-----END PGP SIGNATURE-----

Reply via email to