[PATCH] refactorize interface of Curl_ssl_recv/Curl_ssl_send
Date: Fri, 2 Apr 2010 22:52:40 +0200
On Friday 02 of April 2010 20:42:16 Daniel Stenberg wrote:
> On Fri, 2 Apr 2010, Kamil Dudka wrote:
> > Thanks for looking into this! I am about to write the patch and now got
> > a bit confused. What's the proper place for function's documentation in
> > the cURL coding style? It seems fairly inconsistent to me.
> Right, we've never been very picky on how or where the docs for the
> internal functions is made. Inconsistency is far better than total absense!
> > Should the documentation of a function be in .h or in .c when the
> > function has an external declaration? Or both places? If so, should the
> > documentation be identical?
> I think I've tried to put them in both places, but I could also be happy
> with only having them in the .h file.
> > Going through the SSL backends, I see the _send and _recv functions has
> > also some fragments of documentation. What if we replace them by a link
> > to the documentation in sslgen.h or sslgen.c to keep it at one place?
> > /* for documentation see Curl_ssl_send() in sslgen.h */
> Sure, that'd be great!
Here is my attack on interfaces of Curl_ssl_send() and Curl_ssl_recv(), based
on your original patch. I've tried the test-suite with OpenSSL, GnuTLS and
NSS. It looks like nothing is broken by the patch so far. Only tests 301
and 306 fail with GnuTLS on my box, but they fail also without the patch
applied. Any feedbeck welcome!
- text/x-diff attachment: 0001-refactorize-interface-of-Curl_ssl_recv-Curl_ssl_send.patch