ports/64362: Maintainer port update: devel/cppadvio

[oleg at pobox.com]

>Number:         64362
>Category:       ports
>Synopsis:       Maintainer port update: devel/cppadvio, compiling on Alpha
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    freebsd-ports-bugs
>State:          open
>Quarter:        
>Keywords:       
>Date-Required:
>Class:          maintainer-update
>Submitter-Id:   current-users
>Arrival-Date:   Tue Mar 16 22:20:08 PST 2004
>Closed-Date:
>Last-Modified:
>Originator:     oleg at pobox.com
>Release:        FreeBSD 4.9-RELEASE-p1 i386
>Organization:
>Environment:
System: FreeBSD Adric 4.9-RELEASE-p1 FreeBSD 4.9-RELEASE-p1 #1: Tue Dec 30 16:04:05 PST 2003 oleg at Adric:/usr/obj/usr/src/sys/Adric i386

>Description:
CPPADVIO fails to build under Alpha:
	http://bento.freebsd.org/errorlogs/alpha-4-latest/cppadvio-2.6_1.log
Correspondingly, devel/cppadvio/Makefile was patched from version 1.7
to version 1.8 to indicate that the port "Does not compile on alpha 4.x"

Alas, I have no access to Alpha. However, I understood
the problem from the error log and am quite sure the submitted patch will
fix it. I tested that the patch doesn't break the i386 4.x port.

Again, I believe that the patch fixes the compilation problem on
Alpha, but I have no way to prove it. I would greatly appreciate it
indeed if a kind committer could try the patch on Alpha. If it works,
could you also revert devel/cppadvio/Makefile from version 1.8 to
version 1.7 (so the port would be automatically built on Alpha). The patch
does seem to work on i386, and it makes the code more portable.

Thank you!

>How-To-Repeat:
>Fix:

New file /usr/ports/devel/cppadvio/files/patch-TCPstream.cc
--- patch begins here ---
--- TCPstream-orig.cc	Tue Nov 14 19:06:44 2000
+++ TCPstream.cc	Tue Mar 16 21:34:58 2004
@@ -44,7 +44,7 @@
 {
                                 // Convertion between the host and the network
                                 // byte orders
-#if !defined(htons) && !defined(linux)
+#if !defined(htons) && !defined(linux) && !defined(_MACHINE_ENDIAN_H_)
 unsigned short htons(unsigned int data);        // For a short data item
 unsigned short ntohs(unsigned int data);        // For a short data item
 unsigned long  htonl(unsigned long data);       // For a long data item
--- patch ends here ---
>Release-Note:
>Audit-Trail:
>Unformatted: