1 |
On Wed, 01 Dec 2010 22:44:45 +0100 |
2 |
Diego Elio Pettenò <flameeyes@×××××.com> wrote: |
3 |
|
4 |
> Il giorno mer, 01/12/2010 alle 22.30 +0100, Michał Górny ha scritto: |
5 |
> > |
6 |
> > In other words, the preferred complete build of ekg2 requires both |
7 |
> > gnutls and openssl. However, all of the features should work fine |
8 |
> > with openssl itself as well. |
9 |
> > |
10 |
> There is a simple first question to ask here. Does it use pure GnuTLS |
11 |
> or the GnuTLS-OpenSSL glue layer? If the latter, then you either |
12 |
> build it to only use one, or you're going to hit symbol collisions |
13 |
> which are no fun to debug. |
14 |
|
15 |
Pure GnuTLS. |
16 |
|
17 |
> In general, as much as I like to allow alternatives, I'd suggest you |
18 |
> to pick a stand, and since OpenSSL is the lowest common denominator, |
19 |
> I'd suggest you pick that one. But for sure there will be people |
20 |
> complaining about licensing… |
21 |
|
22 |
Well, I thought about having 'ssl' to enable SSL support through |
23 |
OpenSSL and 'gnutls' to enable GnuTLS for the plugins supporting it. |
24 |
Then, logically, 'openssl' flag would be useful no more but wouldn't |
25 |
that imply exceeding the meaning of the 'ssl' flag? |
26 |
|
27 |
> How feasible would be to get upstream to implement GnuTLS support on |
28 |
> the remaining plugins? |
29 |
|
30 |
For the IRC plugin, it's pretty possible. For the sim plugin, I think |
31 |
it would be much harder. |
32 |
|
33 |
-- |
34 |
Best regards, |
35 |
Michał Górny |