Yang Tse [Wed, 1 Nov 2006 18:33:50 +0000 (18:33 +0000)]
test 518 is all about testing libcurl functionality
when more than FD_SETSIZE file descriptors are open.
This means that if for any reason we are not able to
open more than FD_SETSIZE file descriptors then test
518 should not be run.
test 537 is all about testing libcurl functionality
when the system has nearly exhausted the number of
free file descriptors. Test 537 will try to run with
very few free file descriptors.
Yang Tse [Fri, 27 Oct 2006 15:32:18 +0000 (15:32 +0000)]
Compiler warning fix.
Assigning the const value zero to a pointer to function
results in a null pointer value assignment to the function
pointer.
Assignment of any nonzero value is what should result in a
implementation compiler dependent result.
Since what we want to do here is the first case, this should
not trigger compiler warnings related with conversions from
'pointer to data' to 'pointer to function'.
Yang Tse [Thu, 26 Oct 2006 13:55:24 +0000 (13:55 +0000)]
Improved rlimit logic:
- Take in account RLIM_INFINITY.
- Verify that soft limit is actually changed when doing so.
- Show errno in case getrlimit or setrlimit fails.
- Keep file descriptors open only while runing this test.
Daniel Stenberg [Wed, 25 Oct 2006 20:40:14 +0000 (20:40 +0000)]
Fixed CURLOPT_FAILONERROR to return CURLE_HTTP_RETURNED_ERROR even for the
case when 401 or 407 are returned, *IF* no auth credentials have been given.
The CURLOPT_FAILONERROR option is not possible to make fool-proof for 401
and 407 cases when auth credentials is given, but we've now covered this
somewhat more.
You might get some amounts of headers transferred before this situation is
detected, like for when a "100-continue" is received as a response to a
POST/PUT and a 401 or 407 is received immediately afterwards.
Yang Tse [Tue, 24 Oct 2006 15:51:42 +0000 (15:51 +0000)]
Abort test if it seems that it would have run forever. This is just to prevent
test hanging and actually is an indication that there's a condition that is
not being properly handled at some point in the library.
Remove a pair of braces and adjust indentation appropriately.
Daniel Stenberg [Sat, 21 Oct 2006 11:40:04 +0000 (11:40 +0000)]
Nir Soffer made the tests/libtest/Makefile.am use a proper variable for all
the single test applications' link and dependences, so that you easier can
override those from the command line when using make.
Daniel Stenberg [Sat, 21 Oct 2006 11:32:05 +0000 (11:32 +0000)]
Armel Asselin separated CA cert verification problems from problems with
reading the (local) CA cert file to let users easier pinpoint the actual
problem. CURLE_SSL_CACERT_BADFILE (77) is the new libcurl error code.
Yang Tse [Fri, 20 Oct 2006 17:16:06 +0000 (17:16 +0000)]
Since now src/setup.h includes setup_once.h, src/config-win32.h needs
the definitions for the return type and arguments types of functions
recv() and send().
Yang Tse [Fri, 20 Oct 2006 15:39:54 +0000 (15:39 +0000)]
Decrease the posibility of aborting a test which actually is not
stale by replacing loop counters with timeouts. In this way the
main loop of the test will be allowed to run up to 30 seconds on
any platform before aborting it.
Daniel Stenberg [Fri, 20 Oct 2006 12:25:39 +0000 (12:25 +0000)]
When a resolve is made on a pipelined connection we need to detect it properly
(when the resoling isn't completede yet) and not confuse it with a simple
connection re-use (non-pipelined).
Yang Tse [Thu, 19 Oct 2006 17:29:25 +0000 (17:29 +0000)]
Abort test if it seems that it would have run forever. This is just to prevent
test hanging and actually is an indication that there's a condition that is
not being properly handled at some point in the library.
Loop counter limits might need to be further increased on false positives.
Daniel Stenberg [Thu, 19 Oct 2006 14:28:50 +0000 (14:28 +0000)]
Here's an effort to avoid saying 'data not shown' in the debug parts when the
data is actually shown on screen. Like when you do 'curl -v host' with data
and debug info sent to the same terminal.
Daniel Stenberg [Wed, 18 Oct 2006 15:10:49 +0000 (15:10 +0000)]
When a connection is re-used, it can be flagged for re-use before the name
resolving is completed so we must make sure to survive it and mark the
connection as async (ie not yet connected completely).