Subject: Re: [PATCH] Added Windows Cryptography API: Next Generation backend

Re: [PATCH] Added Windows Cryptography API: Next Generation backend

From: Marc Hoersken <>
Date: Tue, 18 Mar 2014 22:47:53 +0100

Hi Bob,

On 17.03.2014 17:42, Bob Kast wrote:
> Now that WinCNG is in, should I post my patches on top of that?

Yes, please post your patches rebased on the current repository head,
e.g. master.

> The VS files in the repository are .dsw/.dsp files. These are for a very old
> version of VS and I doubt are very useful except for a starting point for
> current VS support. I have not changed these files so there should not be a
> backward compatibility issue. The VS files I added (.sln/.vcxproj) do make
> some assumptions:
> - Uses WinCNG
> - Uses MSVCRT*.DLL. This implies that these have been "installed" somehow.

Ah, I am sorry, I must have had a dirty working directory before merging
your patch. I somehow thought the changes to the makefiles were part of
your patch but they might actually have been a modification I did to my
local directory a while ago.

> I could add a lot more configurations for each permutation but that might
> just add to the confusion. If you think the above assumptions don't
> represent the most widely used configuration, I could certainly change them.
> I would think some doc file that describes how to change the config might be
> the best way to go.

As far as I can tell the VS files are currently automatically generated
using the Makefile in the root directory. This happens within the dist
target which is used to build distributable tar bundles of libssh2.
Maybe the updated VS makefiles could also be automatically generated
from the existing build infrastructure in order to reduce the amount of
redundant build configuration. I usually handcraft the VS project files
for my own projects after having VS setup the initial components.

> Style question: VS logs many warnings due to "possible data loss" (and
> others), such as assigning a long to a short. This is throughout the
> library, not just in WinCNG. These can be 'fixed' easily with casts, less
> easily by matching data types, or I can just tell the compiler to not check
> this. Should I post "fixes" to these warnings or just ignore them?

I would suggest posting individual patches grouped by issue, so that the
maintainers can judge them and merge them if they are actually a problem
and not just a false-positive.

Best regards,
Received on 2014-03-18