Re: [PATCH/RFC] http_init: only initialize SSL for https

On Mon, Mar 18, 2013 at 11:38 AM, Erik Faye-Lund [off-list ref] wrote:

On Sun, Mar 17, 2013 at 11:27 PM, Junio C Hamano [off-list ref] wrote:

quoted

Daniel Stenberg [off-list ref] writes:

quoted

On Sun, 17 Mar 2013, Antoine Pelisse wrote:

quoted

quoted

With redirects taken into account, I can't think of any really good way
around avoiding this init...

Is there any way for curl to initialize SSL on-demand ?

Yes, but not without drawbacks.

If you don't call curl_global_init() at all, libcurl will notice that
on first use and then libcurl will call global_init by itself with a
default bitmask.

That automatic call of course will prevent the application from being
able to set its own bitmask choice, and also the global_init function
is not (necessarily) thread safe while all other libcurl functions are
so the internal call to global_init from an otherwise thread-safe
function is unfortunate.

So in short, unless you are writing a custom application to talk to
servers that you know will never redirect you to HTTPS, passing
custom masks such as ALL&~SSL to global-init is not going to be a
valid optimization.

I think that is a reasonable API; your custom application may want
to go around your intranet servers all of which serve their status
over plain HTTP, and it is a valid optimization to initialize the
library with ALL&~SSL.  It is just that such an optimization does
not apply to us---we let our users go to random hosts we have no
control over, and they may redirect us in ways we cannot anticipate.

I wonder. Our libcurl is build with "-winssl" (USE_WINDOWS_SSPI=1), it
seems. Perhaps switching to openssl (which we already have libraries
for) would make the init-time better?

It does indeed. So this is probably a better solution, and is
something we're considering doing in Git for Windows anyway (for a
different reason). Thanks for all the feed-back!