On Aug 20, 2014, at 2:24 AM, Mark Evenson evenson@panix.com wrote:
Even though we are passing all the CFFI tests,
CL-USER> (drakma:http-request “https://google.com%E2%80%9D)
still times out with an error. Always something further to fix…
Yes, I’ve run across this too. The first https call hangs, even if one adds :connection-timeout 120. Could we be improperly initializing something somewhere along the run? After recovering from this error, subsequent calls work. This is probably more of a cl+ssl issue than a cffi (or ABCL) issue, but it definitely seems to be ABCL-specific.
Thanks for your work to clean up the CFFI patch set! I look forward to the day when I can run stock CFFI, cl+ss, and ABCL and have all my code working :)
Cyrus
On 21 Aug 2014, at 01:55, Anton Vodonosov avodonosov@yandex.ru wrote:
What version of cl+ssl do you test?
On 2014-07-19 a pull request was merged int cl+ssl. According to the author, https works OK (mostly) for him. https://github.com/cl-plus-ssl/cl-plus-ssl/pull/5
As Anton notes in the followup, the author of CL+SSL pull request #5 was Olof. After visual inspection of CL+SSL #5, it seems that Olof might be relying [on the one patch that didn’t make it as part of my CFFI pull request][1], as it wasn’t readily apparent to me how to merge.
This message constitutes a “HEADS UP” to remind myself (and possibly others) to cross-check what exactly the behaviour of CFFI #43 is with CL-SSL #5, and to revise CFFI #43 if necessary.
[1]: https://github.com/easye/cffi/commit/36012f655372217133b2a83fb3008380724cb37...
armedbear-devel@common-lisp.net
-
Anton Vodonosov -
Cyrus Harmon -
Mark Evenson