Daniel Stenberg [Tue, 14 Aug 2001 08:24:37 +0000 (08:24 +0000)]
corrected the size_t weirdness. Expect size_t to be unsigned. Moved most
over to ssize_t that is signed. Removed all the special-purpose VMS #ifdefs
that were added for this.
Daniel Stenberg [Tue, 14 Aug 2001 08:19:40 +0000 (08:19 +0000)]
Added an empty win32_cleanup for non-windows systems to prevent compiler
warnings, changed the Curl_open() call as the second argument was never
used anyway
Daniel Stenberg [Tue, 14 Aug 2001 08:18:35 +0000 (08:18 +0000)]
removed the *done() function as it served no purpose, added type casts when
converting from 'const char *' to 'char *' to please my picky compiler
options
Daniel Stenberg [Fri, 10 Aug 2001 06:24:49 +0000 (06:24 +0000)]
moved the download/upload speed calculations, to be made on every invoke
of the progressupdate, as on very quick transfers they wouldn't always get
calculated!
Daniel Stenberg [Thu, 9 Aug 2001 09:47:53 +0000 (09:47 +0000)]
The redirected error stream was closed before curl_easy_cleanup() was made,
and when VERBOSE was enabled, that used the stream. Also, the stream was
closed even if we looped to get more files.
Corrects Dustin Boswell's bug report #441610
Daniel Stenberg [Wed, 8 Aug 2001 07:35:57 +0000 (07:35 +0000)]
The file name given to -E can now contain drive letters on windows, if they
start the file name as in 'X:\' where X is any letter. The colon otherwise
normally separate the file name from the password.
Daniel Stenberg [Mon, 6 Aug 2001 12:47:39 +0000 (12:47 +0000)]
Nico's fixes for VMS, most of these are fixes for bad uses of size_t that
forgets that it is very often unsigned. These should be fixed globally and
then many #ifdef VMS lines can be removed.
Daniel Stenberg [Mon, 6 Aug 2001 12:36:18 +0000 (12:36 +0000)]
Curl_FormFree renamed to Curl_formclean, as it turns out VMS for example
requires all global symbols to be *case insentively* unique! curl_formfree
is a global function we shouldn't touch.
Daniel Stenberg [Mon, 6 Aug 2001 08:22:26 +0000 (08:22 +0000)]
Jonathan Hseu noticed that you couldn't get a header callback unless you
set CURLOPT_WRITEHEADER to non-NULL, even if you didn't care about that
data. This is now fixed.
Daniel Stenberg [Mon, 6 Aug 2001 08:18:15 +0000 (08:18 +0000)]
corrected the comment for CURLOPT_WRITEHEADER in setopt(), and made it
read a void * and not a FILE *, as that was how it used to work and not
anymore...
Daniel Stenberg [Sun, 5 Aug 2001 12:30:57 +0000 (12:30 +0000)]
- Sergio Ballestrero provided a patch for reading responses from NCSA httpd
1.5.x servers, as they return really screwed up response headers when asked
for with HTTP 1.1.
Add win32 initialization support to curl_global_init() and
curl_global_cleanup(). Update corresponding man pages...
Improve the logic in curl_global_cleanup() and curl_global_init() so that
they are not called twice if the application libraries have been
initialized and make sure to reset the init flags in curl_global_cleanup().
Daniel Stenberg [Mon, 25 Jun 2001 12:56:06 +0000 (12:56 +0000)]
'resume from'-fix. When using "-C -" and doing multiple file download, the
first resume was used on all files instead of being reset between each file.
As found by Björn Stenberg.
Daniel Stenberg [Wed, 20 Jun 2001 07:43:36 +0000 (07:43 +0000)]
when --with-ssl is used with a specified path, we use that path immediately
to check libs and include files in, we don't check the default places first!