Dynamic OpenSSL import library

[Tim Burrell]

Hi all,

I'm one of TransGaming's R&D developers working on our portability products.

We've recently done a cleanup and refactoring to the OpenSSL loading 
routines; rather than loading the libraries in a few different places we 
moved the code into an OpenSSL importer that consolidates the loading 
into a separate library.

Our importer does some additional things that might not be overly useful 
to Wine, such as detecting and loading different versions of the OpenSSL 
libraries at runtime (rather that using the version detected while 
configuring), but the cleanup and refactoring worked out quite well, and 
we think the code would still be valuable to Wine.

We could port this code back to the Wine tree, but it will take some 
effort to do so.  Also, our current implementation of the dynamic 
OpenSSL importer isn't complete; we're only loading the specific 
functions we've needed up until now.

Before going to the effort of completing the library, and porting it 
back to the Wine tree, I thought I'd reach out first to see if it's even 
something that's of interest?  If it is, great, we can complete the 
library and submit some patches for review, but if the concept isn't of 
interest then we'll just continue using it internally.

I've attached a copy of the main C file from the OpenSSL import library, 
so you can get an idea for what it will look like.  Keeping in mind that 
it's not finished yet, and needs some cleaning up, does this approach 
look acceptable?  Also attached is a patch (that won't directly apply) 
that should give an idea of its intended usage.

Anyone have thoughts / questions?

Thanks,

Tim.

[Henri Verbeet]

On 25 February 2011 17:56, Tim Burrell  wrote:
> Anyone have thoughts / questions?

I'm not sure how relevant the problem this solves is to Wine, it's
strikes me as something that's mainly interesting for people
distributing non-distro binaries.

[Juan Lang]

Hi Tim,
> Before going to the effort of completing the library, and porting it back to
> the Wine tree, I thought I'd reach out first to see if it's even something
> that's of interest?  If it is, great, we can complete the library and submit
> some patches for review, but if the concept isn't of interest then we'll
> just continue using it internally.

I believe we're not interested.  Thanks for offering.  Long term, our
intent is to avoid loading OpenSSL, especially in wininet/winhttp.
wininet and winhttp should use schannel rather than loading OpenSSL
directly.

Thanks,
--Juan