aboutsummaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2003-06-11when we get the auth headers, we still need to read out the full body responseDaniel Stenberg
as otherwise we can re-send requests on the same connection nicely
2003-06-11correct mistakesDaniel Stenberg
2003-06-11describe the NTLM mechanism tooDaniel Stenberg
2003-06-11basic NTLM supportDaniel Stenberg
2003-06-11ntlm addedDaniel Stenberg
2003-06-11Initial take at NTLM authentication. It doesn't really work at this pointDaniel Stenberg
but the infrastructure is there.
2003-06-10clarify the CUSTOMREQUEST and HTTPHEADER options slightlyDaniel Stenberg
2003-06-10CURLOPT_HTTPAUTH docuDaniel Stenberg
2003-06-10Set auth type differently, we use one CURLOPT_HTTPAUTH instead as we planDaniel Stenberg
to add more method in the future.
2003-06-10Daniel Kouril for the HTTP negotiate stuffDaniel Stenberg
2003-06-10Daniel Kouril's patch that adds HTTP negotiation support to libcurl wasDaniel Stenberg
added.
2003-06-10we fix moreDaniel Stenberg
2003-06-10more genericDaniel Stenberg
2003-06-10Pass the error stream pointer to the URL globber, so that it can reportDaniel Stenberg
errors correctly to the user, if need be. Also fixed so that a missing ] in the globbing process no longer leads to core dump.
2003-06-06When doing very big GET requests over HTTPS, we need to add some extraDaniel Stenberg
funky logic in order to make re-tries work fine with OpenSSL. This corrects the problem David Orrell noticed.
2003-06-06Just moved around some logic in Curl_write() to make it easier to debug.Daniel Stenberg
2003-06-06spellfixDaniel Stenberg
2003-06-05Reversed the logic to only include the <sys/select.h> header on systemsDaniel Stenberg
known to really NEED it as another system that doesn't have it came up: very old Linux libc5-based systems (as addition to all HPUX versions). The only known system at this point is AIX.
2003-06-03LDAP problem added as mention in bug report #735752Daniel Stenberg
2003-06-03include the time headers just like we used to do in the curl/curl.h headerDaniel Stenberg
once upon the time
2003-06-03we want the time defines tooDaniel Stenberg
2003-06-03Content-Length: now overrides other means of knowing when the stream hasDaniel Stenberg
ended.
2003-06-02Make the Content-Length info override the Connection: close header, so thatDaniel Stenberg
libcurl will stop reading when the number of bytes have arrived and not wait for a closed socket.
2003-06-02the 500-599 test case rangeDaniel Stenberg
2003-06-02ignore moreDaniel Stenberg
2003-06-02ignoreDaniel Stenberg
2003-06-02David Byron's fix to get the progress-bar use the local size too whenDaniel Stenberg
doing a resumed download.
2003-06-02makefile fiddleDaniel Stenberg
changed how http requests are sent - now in one chunk more often HPUX include fix in the external headers better SSL work-arounds for bad SSL servers modified error message when CURLE_HTTP_RETURNED_ERROR is returned
2003-06-02work-around SSL implementation flaws better, pointed out in bug reportDaniel Stenberg
#745122.
2003-06-02make a more descriptive error message when CURLE_HTTP_RETURNED_ERROR isDaniel Stenberg
returned
2003-05-28haven't updates this in a loooong timeDaniel Stenberg
2003-05-28Posting static data using POST and chunked encoded now also appends theDaniel Stenberg
data to the initial request buffer, if the total post data is less than 100K.
2003-05-27include sys/time.h, we didn't have a time() proto anymore. Did one of theDaniel Stenberg
changes in curl/curl.h make this occur?
2003-05-27Documented which rules the public headers must follow when we writeDaniel Stenberg
preprocessor checks for condititions.
2003-05-27oops, removed a # too manyDaniel Stenberg
2003-05-27remove usage of HAVE_* defines, we cannot and shall not depend on any suchDaniel Stenberg
defines in the public external header files
2003-05-27sys/select.h is not present on HPUX, avoid including itDaniel Stenberg
2003-05-27made it work ;-)Daniel Stenberg
2003-05-27if cvs update fails, attempt again after 5 seconds and retry 50 timesDaniel Stenberg
before giving up
2003-05-27Only build in lib and src by default, make the others dist-subdirs.Daniel Stenberg
Make the test stuff get built when we run 'make test' instead.
2003-05-27Rudy Koento experienced problems with curl's recent habit of POSTing data inDaniel Stenberg
two separate send() calls, first the headers and then the data. I've now made a fix that for static and known content that isn't to be chunked-encoded, everything is now sent in one single system call again. This is also better for network performance reasons.
2003-05-27runs on DOS nowDaniel Stenberg
2003-05-27include file flaw and yet another socks5-fixDaniel Stenberg
2003-05-27Another socks5-fix. Make sure that when we use a socks-proxy, it is not theDaniel Stenberg
same as using a httpproxy so we must make sure to better check for http proxies before we do HTTP proxy stuff. This included authorization and URI usage in the request etc.
2003-05-27CURLOPT_HTTPDIGEST is addedDaniel Stenberg
2003-05-26languageDaniel Stenberg
2003-05-26ftp ASCII transfers in general need fixingDaniel Stenberg
2003-05-26Chris Lewis mentioned that he doesn't get WIN32 defined, only _WIN32 so weDaniel Stenberg
make an adjustment to catch this.
2003-05-23even moreDaniel Stenberg
2003-05-23Ricardo Cadime found a socket leak when listing directories withoutDaniel Stenberg
contents. Test cases 144 and 145 were added to verify the fix. Now we deal with return code 450 properly and other codes also do proper cleanup.