Could someone with SSL module core knowledge take a look at SSL.handshake.server_derive_master_secret? In the KE_dh_anon case in the main switch premaster_secret is set to dh_state->get_shared(). Since premaster_secret is a string and get_shared returns a Gmp.mpz object, this is wrong. What should it be?
In the meantime, maybe you could fix cipher.pike so that it compiles again? "Couldn't find program: constants" isn't a very useful result when trying to load it.
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
Previous text:
2003-01-27 19:13: Subject: KE_dh_anon
Could someone with SSL module core knowledge take a look at SSL.handshake.server_derive_master_secret? In the KE_dh_anon case in the main switch premaster_secret is set to dh_state->get_shared(). Since premaster_secret is a string and get_shared returns a Gmp.mpz object, this is wrong. What should it be?
/ Martin Nilsson (Åskblod)
No, that doesn't sound very useful. However, the cipher file shouldn't be used at all. What code does?
/ Martin Nilsson (Åskblod)
Previous text:
2003-01-27 20:04: Subject: KE_dh_anon
In the meantime, maybe you could fix cipher.pike so that it compiles again? "Couldn't find program: constants" isn't a very useful result when trying to load it.
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
The testsuite tries to load everything. If cipher shouldn't be used, remove it. And what about the "constants.pmod" file? Apparently it doesn't work as you had intended, so maybe it should be removed too?
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
Previous text:
2003-01-27 20:09: Subject: KE_dh_anon
No, that doesn't sound very useful. However, the cipher file shouldn't be used at all. What code does?
/ Martin Nilsson (Åskblod)
It works as intended. Protocols.HTTP doesn't crash, which it does without it.
/ Martin Nilsson (Åskblod)
Previous text:
2003-01-27 20:15: Subject: KE_dh_anon
The testsuite tries to load everything. If cipher shouldn't be used, remove it. And what about the "constants.pmod" file? Apparently it doesn't work as you had intended, so maybe it should be removed too?
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
I thought the intention was that `inherit "(SSL.)constants"' should keep working. If it's not for backwards compatibility, why not simply change Protocols.HTTP to do the right thing instead (whatever that is)?
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
Previous text:
2003-01-27 20:19: Subject: KE_dh_anon
It works as intended. Protocols.HTTP doesn't crash, which it does without it.
/ Martin Nilsson (Åskblod)
No, the intention was to keep 'import SSL.constants' working. The master plan was screwed up, since the strict typing actually found bugs in the module. I would perhaps had made a better intermediate state if that hadn't been the case.
/ Martin Nilsson (Åskblod)
Previous text:
2003-01-27 20:24: Subject: KE_dh_anon
I thought the intention was that `inherit "(SSL.)constants"' should keep working. If it's not for backwards compatibility, why not simply change Protocols.HTTP to do the right thing instead (whatever that is)?
/ Marcus Comstedt (ACROSS) (Hail Ilpalazzo!)
My guess is that you should just add a ->digits(256) to convert it to a string, but one would have to check the spec to be sure.
/ Niels Möller ()
Previous text:
2003-01-27 19:13: Subject: KE_dh_anon
Could someone with SSL module core knowledge take a look at SSL.handshake.server_derive_master_secret? In the KE_dh_anon case in the main switch premaster_secret is set to dh_state->get_shared(). Since premaster_secret is a string and get_shared returns a Gmp.mpz object, this is wrong. What should it be?
/ Martin Nilsson (Åskblod)
pike-devel@lists.lysator.liu.se
-
Marcus Comstedt (ACROSS) (Hail Ilpalazzo!) ＠ Pike (-) developers forum -
Martin Nilsson (�skblod) ＠ Pike (-) developers forum
-
Niels M�ller () ＠ Pike (-) developers forum