[libssh2] trivial patches for AIX and HPUX + c89 reminder

2007-04-26 Thread Peter O'Gorman
Hi, Some of the examples did not build because sys/time.h was needed, I added this to the #ifndef WIN32 block becuase I have no idea if win32 has a sys/time.h :) Also, we need libssh2 to build with older compilers, so something needs to be done for src/session.c. I don't really expect this part o

Re: [libssh2] trivial patches for AIX and HPUX + c89 reminder

2007-04-26 Thread Guenter Knauf
Hi Peter, > Some of the examples did not build because sys/time.h was needed, I > added this to the #ifndef WIN32 block becuase I have no idea if win32 > has a sys/time.h :) please check recent CVS and report back; I've just changed the samples to use configure vars, and they should now pick up

[libssh2] gcc 4.x warings about 'differ in signedness'

2007-04-26 Thread Guenter Knauf
Hi all, when I compile with gcc 4.x I see a couple of 'differ in signedness' warnings: http://www.gknw.net/test/libssh2/build2.log I can suppress the warnings with the gcc 4.x-only -Wno-pointer-sign switch: http://www.gknw.net/test/libssh2/build1.log are we going to fix these soon? And if not - i

Re: [libssh2] gcc 4.x warings about 'differ in signedness'

2007-04-26 Thread Dan Fandrich
On Fri, Apr 27, 2007 at 02:49:59AM +0200, Guenter Knauf wrote: > Hi all, > when I compile with gcc 4.x I see a couple of 'differ in signedness' warnings: > http://www.gknw.net/test/libssh2/build2.log > I can suppress the warnings with the gcc 4.x-only -Wno-pointer-sign switch: > http://www.gknw.net

[libssh2] advisability of using struct stat in libssh2_scp_recv()

2007-04-26 Thread Peter O'Gorman
Hi, So, we built curl with libssh2 from yesterday, then tried to copy a 95MB file from remote linux machine to local solaris machine. While libssh2 correctly found the size of the file, curl thought it was a 4492PB file. Yes, that's 4492 petabytes. The reason for this is fairly simple, curl has

Re: [libssh2] gcc 4.x warings about 'differ in signedness'

2007-04-26 Thread Daniel Stenberg
On Thu, 26 Apr 2007, Dan Fandrich wrote: > I get a lot of signed/unsigned comparison warnings on an older gcc as well. > libssh2's use of signed vs. unsigned variables seems pretty inconsistent; > unfortunately, that extends to the public API as well which is more > difficult to change. I thin

Re: [libssh2] advisability of using struct stat in libssh2_scp_recv()

2007-04-26 Thread Daniel Stenberg
On Thu, 26 Apr 2007, Peter O'Gorman wrote: > I suggest using a libssh2_stat structure that will not change size or > offsets with compiler options. I completely agree! - This SF.net email is sponsored by DB2 Express Downloa