On 03.11.21 11:16, Daniel Gustafsson wrote:
>> On 30 Oct 2021, at 14:11, Peter Eisentraut <peter.eisentraut@enterprisedb.com> wrote:
>>
>> On 24.08.21 11:13, Peter Eisentraut wrote:
>>> So I'm tempted to suggest that we remove the built-in, non-OpenSSL cipher and hash implementations in pgcrypto
(basicallyINT_SRCS in pgcrypto/Makefile), and then also pursue the simplifications in the OpenSSL code paths described
in[0].
>>
>> Here is a patch for this.
>
> This patch doesn't work on Windows, which I think is because it pulls in
> pgcrypto even in builds without OpenSSL. Poking at that led me to realize that
> we can simplify even more with this. The conditonal source includes can go
> away and be replaced with a simple OBJS clause, and with that the special hacks
> in Mkvcbuild.pm to overcome that.
>
> Attached is a diff on top of your patch to do the above. I haven't tested it
> on Windows yet, but if you think it's in the right direction we'll take it for
> a spin in a CI with/without OpenSSL.
Here is a consolidated patch. I have tested it locally, so it should be
okay on Windows.
> Now, *if* we merge the NSS patch this does introduce special cases again which
> this rips out. I prefer to try and fix them in that patch to keep avoiding the
> need for them rather than keep them on speculation for a patch which hasn't
> been decided on.
Okay, I wasn't sure about the preferred way forward here. I'm content
with the approach you have chosen.