The latest cut from CVS (as of this morning at 11) builds libcryptopp.lib under gcc 3.2.2 without issue (a couple warnings about deprecated headers, but nothing of real immediate importance). However, test.cpp appears to be missing an include. Around line 36, inside the #ifdef for Berkeley sockets, I needed to add : "#include <netinet/tcp.h>" to get it to compile. This was doing:
gmake clean
gmake all
from the command line. Is there some path setup that might be erroneous on my system to cause that or is it a difference in the std headers? Is that a difference with Cygwin and Gcc?


The rest built fine and "cryptest v" passed all tests.

HTH.

michael


At 07:58 PM 6/20/2003 -0400, you wrote:
I'm able to compile the latest code in CVS using cygwin/GCC 3.2.
Unfortunately I don't have access to a Linux machine with GCC 3.2 so I
can't test it on Linux. The SourceForge compile farm only seems to have
2.9x.

Is there a public Linux shell server with GCC 3.2 installed, or can
someone create a private account for me on their machine?

On Mon, Jun 16, 2003 at 03:03:20PM -0700, Michael Hunley wrote:
> A few people have posted, including a previous post from me, on linker
> errors from the latest version in CVS using GCC under RedHat linux (I use 8
> and GCC 3.2.2, unsure of my lib versions at the moment). The last post
> about it I saw was about 1 week ago. I have not seen any posts discussing
> solutions (though a solution by Wei on the AIX end might be part of it),
> only questions about it. Is anyone working on it? Has it been fixed? I
> have not had the chance to retry the latest cut from CVS, let alone work on
> it. I first noticed the problem after some "code patches" went in to
> optimize the linker for space, under Dev Studio. I do not know if the GCC
> linker issues are related or it was just happenstance. I plan to try
> building it again soon, but was hoping someone knew the state before I
> wasted any time on it.
>
> thanks.
>
> Michael Hunley
> Senior Engineer
> PocketPurchase, Inc.



Reply via email to