On Fri, 14 Dec 2001, Götz Babin-Ebell wrote:

> Today I had some time to look in my patch for OpenSSL ENGINE. According to
> the comments from Daniel I did some changes:

I like this version a lot more. This is near perfect, I'd say. Currently I
can only see one little detail that I'd like your comment on before I go
ahead and apply the lot:

> - case CURLOPT_SSLCERTPASSWD:

You remove this option. That'll make all programs that use this to suddenly
fail when this has been applied.

This option seems to have been replaced with the CURLOPT_SSLKEYPASSWD option.
Is there any reason why we can't accept both to set the password for the
private key? The password isn't required for the certificate anyway, is it? I
mean, isn't the option badly named in the first place?

Another little nit:

You have some error-reporting conditional on the preprocessor symbol DEBUG.
Without that, there's no "plain text" error reported. You should set one with
failf() in the same manner as other code. Humans appreciate error texts! ;-)

Do you have any little test source code that uses this? It would be a really
swell addition to the docs/examples section... We will also need all options
documented in docs/curl_easy_setopt.3, but we don't have to do it all at
once...

-- 
    Daniel Stenberg -- curl groks URLs -- http://curl.haxx.se/