Re: patch - Initial set of files for Curl on VMS using GNV toolkit.

2013-07-23 Thread John E. Malmberg
On 7/15/2013 12:34 AM, John E. Malmberg wrote: This is the initial set off files for building Curl on OpenVMS using the GNV (GNu on VMS) tool kit, and the port of Bash 4.2.45 for OpenVMS that will be released shortly. Any feedback? I will need to revise the VMS Readme. file for the side effect

Re: Problem with curl_formadd()

2013-07-23 Thread Byrial Jensen
Den 23-07-2013 22:17, Daniel Stenberg skrev: On Tue, 23 Jul 2013, Byrial Jensen wrote: I have a problem with curl_formadd() which I suspect is a bug: fails with CURLcode 2 (CURLE_FAILED_INIT) which I do not understand. First, this is a minor mistake. curl_formadd() does not return the gene

Re: Problem with curl_formadd()

2013-07-23 Thread Daniel Stenberg
On Tue, 23 Jul 2013, Byrial Jensen wrote: I have a problem with curl_formadd() which I suspect is a bug: fails with CURLcode 2 (CURLE_FAILED_INIT) which I do not understand. First, this is a minor mistake. curl_formadd() does not return the general libcurl error code! It returns a CURLFORM

Re: CURLOPT_READFUNCTION performance issue

2013-07-23 Thread Michael Dowling
Hi Daniel, Thanks for your reply and looking into the issue! I apologize if this doesn't continue the email thread properly on the list. I accidentally signed up for the digest mail, but I've updated my subscription now :) > So this version pair on Linux has problems while not on Mac? And if you

Problem with curl_formadd()

2013-07-23 Thread Byrial Jensen
Hi, I am writing a bot to edit www.wikidata.org using libcurl. I have a problem with curl_formadd() which I suspect is a bug: My statement of the form curl_formadd (&post, &last, CURLFORM_PTRNAME, name, CURLFORM_FILECONTENT, "file.txt", CURLFORM_END) fails with CURLcode 2 (C

Re: unpausing an upload

2013-07-23 Thread Justin Karneges
On Tue, Jul 23, 2013 at 11:26 AM, Daniel Stenberg wrote: > On Tue, 23 Jul 2013, Justin Karneges wrote: > > I suggest a fix like the attached one, what do you think about it? >>> >> >> Doesn't seem to work. I expect that my timer callback should get called >> during curl_easy_pause() to set up a

Re: Flawed dotdot removal from path when using an HTTP proxy

2013-07-23 Thread Daniel Stenberg
On Tue, 23 Jul 2013, Fabian Keil wrote: I also attached a potential fix, but I suspect someone more familiar with libcurl's internals could come up with a more elegant solution. Finally there's a trivial comment fix for dotdot.c. Thanks, I applied and merged them all! -- / daniel.haxx.se

Re: unpausing an upload

2013-07-23 Thread Daniel Stenberg
On Tue, 23 Jul 2013, Justin Karneges wrote: I suggest a fix like the attached one, what do you think about it? Doesn't seem to work. I expect that my timer callback should get called during curl_easy_pause() to set up a new timer? No call to my callback is made. I sprinkled some debug prints

Re: multi_socket POST doesn't finish, but PUT does

2013-07-23 Thread Daniel Stenberg
On Mon, 22 Jul 2013, Justin Karneges wrote: I am using POST by setting CURLOPT_POST to 1, setting CURLOPT_POSTFIELDSIZE_LARGE to the request body size, and streaming the input using a read function (and streaming the response with a write function). The entire request gets sent and as far as I

Re: CURLOPT_READFUNCTION performance issue

2013-07-23 Thread Daniel Stenberg
On Mon, 22 Jul 2013, Michael Dowling wrote: I've noticed that there appears to be a significant performance hit when using CURLOPT_READFUNCTION. This issue seems to be platform dependent as I've only been able to get poor performance on Linux (Amazon Linux m1.large 64-bit ami-0358ce33) across

Re: unpausing an upload

2013-07-23 Thread Justin Karneges
On Tue, Jul 23, 2013 at 4:39 AM, Daniel Stenberg wrote: > On Mon, 22 Jul 2013, Justin Karneges wrote: > > Instead of curl_multi_socket_action() with CURL_SOCKET_TIMEOUT, it seems >> the better answer is curl_multi_socket_all(). >> > > Right, that's a better work-around but if you have many handl

Re: unpausing an upload

2013-07-23 Thread Daniel Stenberg
On Mon, 22 Jul 2013, Justin Karneges wrote: Instead of curl_multi_socket_action() with CURL_SOCKET_TIMEOUT, it seems the better answer is curl_multi_socket_all(). Right, that's a better work-around but if you have many handles running that'll cause you quite a significant performance hit. I

Flawed dotdot removal from path when using an HTTP proxy

2013-07-23 Thread Fabian Keil
Test 1231 doesn't work when using an external HTTP proxy. The attached test 1232 reproduces the problem without requiring an external HTTP proxy: fk@r500 ~/git/curl/tests $./runtests.pl -a -n 1232 * System characteristics * curl 7.32.0-DEV (amd64-unknown-freebsd10.0) * libcurl/7

zurl uses libcurl

2013-07-23 Thread Justin Karneges
Alright I *think* I've managed to resolve all my libcurl issues. Zurl is an HTTP client daemon. You send ZeroMQ messages to it and it will preform HTTP requests. Previously the project was using QNetworkAccessManager for outbound HTTP requests. Now it uses libcurl. Source: https://github.com/fano