Am 26.09.19 um 15:41 schrieb Simon Cruanes:
> I've been working on higher level bindings to cURL (based on ocurl), at
> https://github.com/c-cube/ezcurl .
>
> ```
> $ dune utop src/sync
> # Ezcurl.get ~url:"https://ocaml.org/feed.xml" ();;
> - : (Ezcurl_core.response, Curl.curlCode * string) result =
> Ok
>  {Ezcurl_core.code = 200;
>   headers =
>    [("Content-Type", "application/xml"); ("Accept-Ranges", "bytes");
>     ("ETag", "\"1394832364\"");
>     ("Last-Modified", "Thu, 26 Sep 2019 13:15:03 GMT");
>     ("Content-Length", "402487");
>     ("Date", "Thu, 26 Sep 2019 13:49:00 GMT");
>     ("Server", "lighttpd/1.4.53")];
>   body =
>    "<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<feed xmlns=\"http://www.w3.org/2005/Atom\"><icon>http://ocaml.org/img/colour-icon-170x148.png</icon><generator uri=\"https://ocaml.org/\" version=\"${version:syndic}\">OCaml Syndic.Atom feed aggregator</generator><id>urn:md5:302d717d0937d8"... (* string length 402487; truncated *);
>   info =
>    {Ezcurl_core.ri_response_time = 6.346256; ri_redirect_count = 0}}
> ```
>
> Maybe it's a different version of GNUTLS though (I'm on archlinux, it's 3.6.9-1).

Don't think so, because gnutls-cli can actually connect to ocaml.org. I
rather guess it's a protocol option that's creating the problems, in
particular one that changes the order of reads and writes on the socket
(so it's visible to the wrapping library). BTW, it's not only ocaml.org
but e.g. also google.com.

Gerd

-- 
------------------------------------------------------------
Gerd Stolpmann, Darmstadt, Germany    gerd@gerd-stolpmann.de
My OCaml site:          http://www.camlcity.org
Contact details:        http://www.camlcity.org/contact.html
Company homepage:       http://www.gerd-stolpmann.de
------------------------------------------------------------