Theo Buehler
Theo Buehler
This was fixed in OpenBSD's LLVM by adjusting the target info to take our malloc alignment into account, which effectively changes `__STDCPP_DEFAULT_NEW_ALIGNMENT__` to be 16 on i386: https://github.com/openbsd/src/commit/dc1895c3dde03d61ceaa07854da438264dccc5d6 Maybe FreeBSD...
There is no need to expose it only for a test. Once you have a canonized `IPAddrBlocks`, say `addr`, you can use something like ```cmp_func = sk_IPAddressFamily_set_cmp_func(addr, NULL);``` to retrieve...
Of course there had to be a const change in the OpenSSL 1.1 cycle. I should have thought of that. Sorry about hat. Presumably this means that we need a...
On Wed, Oct 25, 2023 at 01:10:27PM -0700, Kazuki Yamaguchi wrote: > Makes sense to me. > > I'm hoping to remove OpenSSL 1.0.2 support in our next release as...
Same thing as in #20250: `penclen` is a `size_t`, so this check isn't correct.
@mspncp Sure, done, see #20435
RFC 3779, section 2.2.3.3: There MUST be only one IPAddressFamily SEQUENCE per unique combination of AFI and SAFI.
A call to [`OPENSSL_init_ssl()`][1] (which is exposed in rust-openssl) with `OPENSSL_INIT_NO_ATEXIT` could solve this. Not sure if it would be acceptable for cryptography to expose this function or to add...
I don't follow. Automatic library initialization is supposed to happen during the first API call that needs it. Why is that different for a system OpenSSL? Once the atexit handler...
On Thu, Jul 06, 2023 at 03:36:02AM -0700, Tormod Volden wrote: > Not of greatest importance, but I put it out here and it would be nice if someone familiar...