To be addressed before 7.19.0 (planned release: August 2008) ============================= 144 - Help apps use 64bit/LFS libcurl! http://curl.haxx.se/mail/lib-2008-07/0120.html Patch (first version) of 64-bit curl_off_t no longer gated to off_t was sent to the mailing list 2008-07-14. After all the changes done afterwards in CVS its sure it doesn't apply clean as is. A second version patch has to be generated. This second patch version should also address the problem the first one has of not working when building outside of the source tree. #148 blocks this. UPDATE: * #148 no longer blocks this. 145 - Phil Blundell's CURLOPT_SCOPE patch/work 146 - Yehoshua Hershberg's re-using of connections that failed with CURLE_PEER_FAILED_VERIFICATION 148 - Introduction of m4/reentrant.m4 is triggering some problems on Solaris systems. The problem manifests when buildconf runs aclocal, at some point aclocal fails when using GNU m4 version 1.4.5 it runs out of memory. If GNU m4 version 1.4.11 is used everything works. Lots of tests have been done which have allowed to discard many possible pitfalls on our side. As of right now my main suspect is which 'sed' binary is being used by aclocal, if it is a broken sed, aclocal might be feeding garbage to m4. As a consecuence of al this tries and tests the inclusion method of m4/reentrant.m4 has been changed several times, and should finally be changed back to the proper '-I m4' method which was used initially and that didn't fail on other systems. http://curl.haxx.se/mail/lib-2008-07/0210.html UPDATE: * sed cannot be the culprit aclocal and autom4te are perl scripts and don't use 'sed'. * autom4te matches autoconf version and aclocal matches automake's. * Even when directly copying into acinclude.m4 the contents of file reentrant.m4 and having the '-I m4' thing disabled the problem persists. * Reinstated the 'aclocal -I m4' in buildconf and 'ACLOCAL_AMFLAGS = -I m4' way of including our local m4/reentrant.m4 file. This even takes care of including the file in the distribution tarball. 149 - The introduction of the sreadfrom macro has proven usefull on all systems except on HP-UX. For this release I will back out its use in lib/tftp.c and test/server/tftpd.c leaving it in the same functional state as it was. The rest of things will be kept in as thay are needed to properly fix it in another release (autoconf work required). UPDATE: * The use of the sreadfrom() wrapper to replace recvfrom() in our code has been undone. * Awaiting autobuild results before updating or closing this issue. 150 - PUT with -C - sends garbage in the Content-Range: header (test case 1041) 151 - PUT with -L hangs after receiving a redirect (test case 1051, but the test harness has a problem with this, too) 152 -