From: Alex P. <pe...@in...> - 2004-09-24 15:00:06
|
Dmitry Yemanov wrote: >"Mark O'Donohue" <mar...@fi...> wrote: > > >>It needs a bit of roadtesting, and a few more eyes, and Im sure a bit >>more polishing, but Alex's design means that: >> >> 1) old client can connect to new servers, >> >> Yes - except gsec. >> 2) new clients still talk to old servers. >> >> Yes - except gsec. It uses new spb parameter to specify security database to work with. >> 3) and even if really needed old security.fdb files could work >> with new servers. >> >> After fix of updatable view in cmp.cpp - yes. I revert back to update of view USERS instead of underlying table in security.epp, and this will work. But it's better to apply script I plan to write to update security database. > >If these points are true, then I support your proposal. Do I understand >correctly that you suggest to have hardcoded two algorithms (MD5/SHA1 + DES) >instead of firebird.conf-specified plugins list proprosed by Alex? Do you >want to add the crypt plugins feature later and remove the hardcoded stuff >or you think it's too complex solution for most hosts? > > > Looking at the number of mentioned in this thread hashing methods I still think that plugins will be useful. Moreover, to be secure we must be able to turn off DES hashes when needed - for example, in default install. Appears configuration parameter will be required for it anyway. Therefore, why not use it at that same time for selecting plugin? One drawback - API for this plugins should be specified. Currently I use extern "C" { FB_DLL_EXPORT char* makehash(const char* salt, const char* password); }; It need not be thread-safe (each plugin call is protected by mutex). Alex. |