platform/upstream/curl.git
15 years agoProper naming for the experimental compiler test and moved to *-compilers.m4
Yang Tse [Thu, 7 May 2009 13:58:15 +0000 (13:58 +0000)]
Proper naming for the experimental compiler test and moved to *-compilers.m4

15 years agoadded Curl_getoff_all_pipelines SIGSEGV
Daniel Stenberg [Thu, 7 May 2009 11:41:06 +0000 (11:41 +0000)]
added Curl_getoff_all_pipelines SIGSEGV

15 years agoadded mention of two more current bugs/issues
Daniel Stenberg [Thu, 7 May 2009 11:20:44 +0000 (11:20 +0000)]
added mention of two more current bugs/issues

15 years agoMoved *_CHECK_COMPILER_HALT_ON_ERROR and *_CHECK_COMPILER_ARRAY_SIZE_NEGATIVE to...
Yang Tse [Thu, 7 May 2009 11:06:48 +0000 (11:06 +0000)]
Moved *_CHECK_COMPILER_HALT_ON_ERROR and *_CHECK_COMPILER_ARRAY_SIZE_NEGATIVE to *-compilers.m4 along with other *_CHECK_COMPILER_*

15 years agoman page warnings fixed
Daniel Stenberg [Thu, 7 May 2009 09:33:13 +0000 (09:33 +0000)]
man page warnings fixed

15 years ago- Man page *roff problems fixed thanks to input from Colin Watson. Problems
Daniel Stenberg [Thu, 7 May 2009 09:31:24 +0000 (09:31 +0000)]
- Man page *roff problems fixed thanks to input from Colin Watson. Problems
  reported in the Debian package.

15 years ago- Vijay G filed bug report #2723236
Daniel Stenberg [Thu, 7 May 2009 08:30:43 +0000 (08:30 +0000)]
- Vijay G filed bug report #2723236
  (http://curl.haxx.se/bug/view.cgi?id=2723236) identifying a problem with
  libcurl's TFTP code and its lack of dealing with the OACK packet.

15 years agoUse autobuilds to verify if a couple of experimental compiler tests pass on all of...
Yang Tse [Wed, 6 May 2009 18:37:24 +0000 (18:37 +0000)]
Use autobuilds to verify if a couple of experimental compiler tests pass on all of them.
This will be removed in 24 or 48 hours.

15 years agocurl_sockaddr struct member addrlen is an 'unsigned int' since 7.18.0
Yang Tse [Wed, 6 May 2009 14:04:05 +0000 (14:04 +0000)]
curl_sockaddr struct member addrlen is an 'unsigned int' since 7.18.0

15 years agofIX *__SOCKLEN_T definitions for OS400 and generic GCC targets
Yang Tse [Wed, 6 May 2009 13:58:41 +0000 (13:58 +0000)]
fIX *__SOCKLEN_T definitions for OS400 and generic GCC targets

15 years agoTemporary and HP-UX autobuild specific change to debug SSH server start-up failures
Yang Tse [Wed, 6 May 2009 10:32:06 +0000 (10:32 +0000)]
Temporary and HP-UX autobuild specific change to debug SSH server start-up failures

15 years agofix compiler warning
Yang Tse [Tue, 5 May 2009 13:04:33 +0000 (13:04 +0000)]
fix compiler warning

15 years agoreintroduce #232
Yang Tse [Tue, 5 May 2009 13:03:22 +0000 (13:03 +0000)]
reintroduce #232

15 years agoMention last changes
Yang Tse [Tue, 5 May 2009 08:52:08 +0000 (08:52 +0000)]
Mention last changes

15 years agoRevert change committed on Mon May 4 09:30:23 2009 UTC.
Yang Tse [Tue, 5 May 2009 08:46:31 +0000 (08:46 +0000)]
Revert change committed on Mon May 4 09:30:23 2009 UTC.

The origin of the problem with test case #251 was the --ftp-port not being the CLIENTIP address.

15 years agoI simplified the SSL session id re-use code now to *always* ditch the previous
Daniel Stenberg [Tue, 5 May 2009 08:33:29 +0000 (08:33 +0000)]
I simplified the SSL session id re-use code now to *always* ditch the previous
one and store the current one, as it makes the code less complex and I'm not
even sure I can check for the same session id using memcmp() like that.

15 years agoFixed the --ftp-port address to the CLIENTIP address.
Yang Tse [Tue, 5 May 2009 08:19:55 +0000 (08:19 +0000)]
Fixed the --ftp-port address to the CLIENTIP address.

15 years ago- Inspired by Michael Smith's session id fix for OpenSSL, I did the
Daniel Stenberg [Mon, 4 May 2009 22:20:09 +0000 (22:20 +0000)]
- Inspired by Michael Smith's session id fix for OpenSSL, I did the
  corresponding fix in the GnuTLS code: make sure to store the new session id
  in case the re-used one is rejected.

15 years ago- Michael Smith posted bug report #2786255
Daniel Stenberg [Mon, 4 May 2009 21:57:14 +0000 (21:57 +0000)]
- Michael Smith posted bug report #2786255
  (http://curl.haxx.se/bug/view.cgi?id=2786255) with a patch, identifying how
  libcurl did not deal with SSL session ids properly if the server rejected a
  re-use of one. Starting now, it will forget the rejected one and remember
  the new. This change was for OpenSSL only, it is likely that other SSL lib
  code needs similar fixes.

15 years agomore return codes added
Daniel Stenberg [Mon, 4 May 2009 21:36:53 +0000 (21:36 +0000)]
more return codes added

15 years agofIX *__SOCKLEN_T definitions for MVS and 370 targets
Yang Tse [Mon, 4 May 2009 17:05:25 +0000 (17:05 +0000)]
fIX *__SOCKLEN_T definitions for MVS and 370 targets

15 years agofIX *__SOCKLEN_T definitions for several Windows target tool-chains
Yang Tse [Mon, 4 May 2009 15:25:57 +0000 (15:25 +0000)]
fIX *__SOCKLEN_T definitions for several Windows target tool-chains

15 years agoa busload symbols added
Daniel Stenberg [Mon, 4 May 2009 14:50:29 +0000 (14:50 +0000)]
a busload symbols added

15 years agoPhilippe Vaucher reported this typo
Daniel Stenberg [Mon, 4 May 2009 13:01:27 +0000 (13:01 +0000)]
Philippe Vaucher reported this typo

15 years agoDavid McCreedy's patch to fix test suite harness to allow test FTP server and client...
Yang Tse [Mon, 4 May 2009 10:30:23 +0000 (10:30 +0000)]
David McCreedy's patch to fix test suite harness to allow test FTP server and client on different machines, providing FTP client address when running the FTP test server.

15 years agoDavid McCreedy's "transfer.c fixes for CURL_DO_LINEEND_CONV and non-ASCII platform...
Yang Tse [Mon, 4 May 2009 09:47:02 +0000 (09:47 +0000)]
David McCreedy's "transfer.c fixes for CURL_DO_LINEEND_CONV and non-ASCII platform HTTP requests" patch

15 years agofix makefile reference to packages/TPF
Yang Tse [Mon, 4 May 2009 00:33:31 +0000 (00:33 +0000)]
fix makefile reference to packages/TPF

15 years agoignore this
Yang Tse [Mon, 4 May 2009 00:31:27 +0000 (00:31 +0000)]
ignore this

15 years agoHP-UX's X/Open network library requirement check follow-up
Yang Tse [Sun, 3 May 2009 19:41:12 +0000 (19:41 +0000)]
HP-UX's X/Open network library requirement check follow-up

15 years agoAdded and disabled test case 563 which shows KNOWN_BUGS #59. The bug
Dan Fandrich [Sun, 3 May 2009 19:25:17 +0000 (19:25 +0000)]
Added and disabled test case 563 which shows KNOWN_BUGS #59.  The bug
report failed to mention that a proxy must be used to reproduce it.

15 years agoHP-UX's X/Open network library requirement check follow-up
Yang Tse [Sun, 3 May 2009 17:35:44 +0000 (17:35 +0000)]
HP-UX's X/Open network library requirement check follow-up

15 years agoThis wasn't intended to be committed.
Yang Tse [Sat, 2 May 2009 03:16:04 +0000 (03:16 +0000)]
This wasn't intended to be committed.

So, until fixed this will remain FIXME.

15 years agoUse build-time configured curl_socklen_t instead of socklen_t
Yang Tse [Sat, 2 May 2009 02:37:32 +0000 (02:37 +0000)]
Use build-time configured curl_socklen_t instead of socklen_t

15 years agoUse build-time configured ares_socklen_t instead of socklen_t
Yang Tse [Sat, 2 May 2009 02:36:47 +0000 (02:36 +0000)]
Use build-time configured ares_socklen_t instead of socklen_t

15 years agoRemove temporary debug tracing for X/Open HP-UX
Yang Tse [Fri, 1 May 2009 22:59:16 +0000 (22:59 +0000)]
Remove temporary debug tracing for X/Open HP-UX

15 years agoMention David McCreedy's patches applied today
Yang Tse [Fri, 1 May 2009 22:55:08 +0000 (22:55 +0000)]
Mention David McCreedy's patches applied today

15 years agoDavid McCreedy's "TPF-platform specific changes to various files" patch follow-up
Yang Tse [Fri, 1 May 2009 22:41:23 +0000 (22:41 +0000)]
David McCreedy's "TPF-platform specific changes to various files" patch follow-up

15 years agofixed: 229 - [PATCH] to correct spelling in ABI
Daniel Stenberg [Fri, 1 May 2009 20:36:39 +0000 (20:36 +0000)]
fixed: 229 - [PATCH] to correct spelling in ABI

15 years agoJose Goncalves fixed my crappy spelling!
Daniel Stenberg [Fri, 1 May 2009 20:20:54 +0000 (20:20 +0000)]
Jose Goncalves fixed my crappy spelling!

15 years ago#2784055, connect race with multi and socks proxy
Daniel Stenberg [Fri, 1 May 2009 20:16:24 +0000 (20:16 +0000)]
#2784055, connect race with multi and socks proxy

bad enough to almost make socks proxies with multi interface unusable

15 years agos/libcurl/c-ares
Daniel Stenberg [Fri, 1 May 2009 18:26:03 +0000 (18:26 +0000)]
s/libcurl/c-ares

15 years agoversion number typo fix
Daniel Stenberg [Fri, 1 May 2009 18:24:34 +0000 (18:24 +0000)]
version number typo fix

15 years agoDavid McCreedy's "http.c fix to Curl_proxyCONNECT for non-ASCII platforms" patch
Yang Tse [Fri, 1 May 2009 13:00:38 +0000 (13:00 +0000)]
David McCreedy's "http.c fix to Curl_proxyCONNECT for non-ASCII platforms" patch

15 years agoDavid McCreedy's "TPF-platform specific changes to various files" patch
Yang Tse [Fri, 1 May 2009 12:39:40 +0000 (12:39 +0000)]
David McCreedy's "TPF-platform specific changes to various files" patch

15 years agomention a bunch of pending patches to deal with
Daniel Stenberg [Fri, 1 May 2009 09:23:14 +0000 (09:23 +0000)]
mention a bunch of pending patches to deal with

15 years agoCheck definition of _XOPEN_SOURCE_EXTENDED with the compiler
Yang Tse [Fri, 1 May 2009 02:03:42 +0000 (02:03 +0000)]
Check definition of _XOPEN_SOURCE_EXTENDED with the compiler

15 years agoAdd temporary debug tracing for X/Open HP-UX
Yang Tse [Thu, 30 Apr 2009 20:33:37 +0000 (20:33 +0000)]
Add temporary debug tracing for X/Open HP-UX

15 years agoCheck if X/Open network library is required
Yang Tse [Thu, 30 Apr 2009 18:35:36 +0000 (18:35 +0000)]
Check if X/Open network library is required

15 years agoWhen generated, display libcurl.pc and libcares.pc
Yang Tse [Thu, 30 Apr 2009 17:06:58 +0000 (17:06 +0000)]
When generated, display libcurl.pc and libcares.pc

15 years agocope with ares_build.h and ares_rules.h follow-up
Yang Tse [Thu, 30 Apr 2009 13:50:44 +0000 (13:50 +0000)]
cope with ares_build.h and ares_rules.h follow-up

15 years agoAdded some notes regarding ares_build.h
Yang Tse [Thu, 30 Apr 2009 12:44:17 +0000 (12:44 +0000)]
Added some notes regarding ares_build.h

15 years agofix EOL
Yang Tse [Thu, 30 Apr 2009 12:12:31 +0000 (12:12 +0000)]
fix EOL

15 years agofix EOL
Yang Tse [Thu, 30 Apr 2009 12:05:53 +0000 (12:05 +0000)]
fix EOL

15 years agocope with ares_build.h and ares_rules.h
Yang Tse [Thu, 30 Apr 2009 11:48:43 +0000 (11:48 +0000)]
cope with ares_build.h and ares_rules.h

15 years agobuildconf.bat for CVS-tree c-ares
Yang Tse [Thu, 30 Apr 2009 10:43:27 +0000 (10:43 +0000)]
buildconf.bat for CVS-tree c-ares

15 years agofix compiler warning: unused parameter
Yang Tse [Thu, 30 Apr 2009 09:32:02 +0000 (09:32 +0000)]
fix compiler warning: unused parameter

15 years ago- I was going to fix issue #59 in KNOWN_BUGS
Daniel Stenberg [Thu, 30 Apr 2009 09:02:39 +0000 (09:02 +0000)]
- I was going to fix issue #59 in KNOWN_BUGS

  If the CURLOPT_PORT option is used on an FTP URL like
  "ftp://example.com/file;type=A" the ";type=A" is stripped off.

  I added test case 562 to verify, only to find out that I couldn't repeat
  this bug so I hereby consider it not a bug anymore!

15 years agoprovide some track record of the SONAME bumps
Daniel Stenberg [Thu, 30 Apr 2009 08:30:52 +0000 (08:30 +0000)]
provide some track record of the SONAME bumps

15 years agomention that cvs is needed for releases
Daniel Stenberg [Thu, 30 Apr 2009 08:08:33 +0000 (08:08 +0000)]
mention that cvs is needed for releases

15 years agoUse 'unsigned int' instead of size_t attempting to avoid header inclusion
Yang Tse [Thu, 30 Apr 2009 01:34:04 +0000 (01:34 +0000)]
Use 'unsigned int' instead of size_t attempting to avoid header inclusion

15 years agoNetWare LibC's getpeername() third argument data type is size_t
Yang Tse [Thu, 30 Apr 2009 00:11:20 +0000 (00:11 +0000)]
NetWare LibC's getpeername() third argument data type is size_t

15 years agoDisplay only CURL_* and CARES_* symbol/macro definitions from curlbuild.h and ares_bu...
Yang Tse [Wed, 29 Apr 2009 19:02:22 +0000 (19:02 +0000)]
Display only CURL_* and CARES_* symbol/macro definitions from curlbuild.h and ares_build.h

15 years agoRemove temporary debug tracing for ares_socklen_t Windows targets
Yang Tse [Wed, 29 Apr 2009 18:08:23 +0000 (18:08 +0000)]
Remove temporary debug tracing for ares_socklen_t Windows targets

15 years agoTrigger c-ares prebuild stage for fake-config setup's
Yang Tse [Wed, 29 Apr 2009 17:57:49 +0000 (17:57 +0000)]
Trigger c-ares prebuild stage for fake-config setup's

15 years agocurl_socklen_t NetWare follow-up
Yang Tse [Wed, 29 Apr 2009 17:07:19 +0000 (17:07 +0000)]
curl_socklen_t NetWare follow-up

15 years agocurl_socklen_t follow-up
Yang Tse [Wed, 29 Apr 2009 15:15:38 +0000 (15:15 +0000)]
curl_socklen_t follow-up

15 years agoares_socklen_t follow-up
Yang Tse [Wed, 29 Apr 2009 15:15:14 +0000 (15:15 +0000)]
ares_socklen_t follow-up

15 years agoares_build.h Windows follow-up
Yang Tse [Wed, 29 Apr 2009 14:05:21 +0000 (14:05 +0000)]
ares_build.h Windows follow-up

15 years agomention the new CURL_SEEKFUNC_ symbols
Daniel Stenberg [Wed, 29 Apr 2009 11:41:29 +0000 (11:41 +0000)]
mention the new CURL_SEEKFUNC_ symbols

15 years ago- Based on bug report #2723219 (http://curl.haxx.se/bug/view.cgi?id=2723219)
Daniel Stenberg [Wed, 29 Apr 2009 11:30:03 +0000 (11:30 +0000)]
- Based on bug report #2723219 (curl.haxx.se/bug/view.cgi?id=2723219)
  I've now made TFTP "connections" not being kept for re-use within libcurl.
  TFTP is UDP-based so the benefit was really low (if even existing) to begin
  with so instead of tracking down to fix this problem we instead removed the
  re-use. I also enabled test case 1099 that I wrote a few days ago to verify
  that this change fixes the reported problem.

15 years agoAdd temporary debug tracing for ares_socklen_t Windows targets
Yang Tse [Wed, 29 Apr 2009 11:11:59 +0000 (11:11 +0000)]
Add temporary debug tracing for ares_socklen_t Windows targets

15 years agoares_build.h NetWare follow-up
Yang Tse [Wed, 29 Apr 2009 09:39:15 +0000 (09:39 +0000)]
ares_build.h NetWare follow-up

15 years agoWhen running testcurl.pl display definitions from ares_build.h
Yang Tse [Wed, 29 Apr 2009 01:31:37 +0000 (01:31 +0000)]
When running testcurl.pl display definitions from ares_build.h

15 years agoallow usage of in-tree c-ares when building from outside of the cvs tree
Yang Tse [Wed, 29 Apr 2009 00:38:08 +0000 (00:38 +0000)]
allow usage of in-tree c-ares when building from outside of the cvs tree

15 years agoares_build.h NetWare attempt
Yang Tse [Wed, 29 Apr 2009 00:17:40 +0000 (00:17 +0000)]
ares_build.h NetWare attempt

15 years ago- Constantine Sapuntzakis filed bug report #2783090
Daniel Stenberg [Tue, 28 Apr 2009 20:27:06 +0000 (20:27 +0000)]
- Constantine Sapuntzakis filed bug report #2783090
  (http://curl.haxx.se/bug/view.cgi?id=2783090) pointing out that on windows
  we need to grow the SO_SNDBUF buffer somewhat to get really good upload
  speeds. http://support.microsoft.com/kb/823764 has the details. Friends
  confirmed that simply adding 32 to CURL_MAX_WRITE_SIZE is enough.

15 years agoallow usage of in-tree c-ares when building from outside of the cvs tree
Yang Tse [Tue, 28 Apr 2009 19:29:50 +0000 (19:29 +0000)]
allow usage of in-tree c-ares when building from outside of the cvs tree

15 years agoInitial step towards a configure time ares_socklen_t definition
Yang Tse [Tue, 28 Apr 2009 16:47:33 +0000 (16:47 +0000)]
Initial step towards a configure time ares_socklen_t definition

15 years agoignore stamp-h*
Yang Tse [Tue, 28 Apr 2009 14:23:28 +0000 (14:23 +0000)]
ignore stamp-h*

15 years agoFixed: 225 - resume when upload from stream, #2709004
Daniel Stenberg [Tue, 28 Apr 2009 11:19:50 +0000 (11:19 +0000)]
Fixed: 225 - resume when upload from stream, #2709004

15 years ago- Bug report #2709004 (http://curl.haxx.se/bug/view.cgi?id=2709004) by Tim
Daniel Stenberg [Tue, 28 Apr 2009 11:19:10 +0000 (11:19 +0000)]
- Bug report #2709004 (curl.haxx.se/bug/view.cgi?id=2709004) by Tim
  Chen pointed out how curl couldn't upload with resume when reading from a
  pipe.

  This ended up with the introduction of a new return code for the
  CURLOPT_SEEKFUNCTION callback that basically says that the seek failed but
  that libcurl may try to resolve the situation anyway. In our case this means
  libcurl will attempt to instead read that much data from the stream instead
  of seeking and that way curl can now upload with resume when data is read
  from a stream!

15 years agoAdded CARES_INCLUDES_SYS_TYPES
Yang Tse [Tue, 28 Apr 2009 10:40:02 +0000 (10:40 +0000)]
Added CARES_INCLUDES_SYS_TYPES

15 years agoRearrange placement inside file of CURL_DEFINE_UNQUOTED, CURL_CONFIGURE_LONG
Yang Tse [Tue, 28 Apr 2009 10:37:06 +0000 (10:37 +0000)]
Rearrange placement inside file of CURL_DEFINE_UNQUOTED, CURL_CONFIGURE_LONG
and CURL_CONFIGURE_CURL_SOCKLEN_T to ease future maintainance.

15 years agoMoved CURL_INCLUDES_INTTYPES to curl-functions.m4 along with other CURL_INCLUDES_*
Yang Tse [Tue, 28 Apr 2009 10:27:04 +0000 (10:27 +0000)]
Moved CURL_INCLUDES_INTTYPES to curl-functions.m4 along with other CURL_INCLUDES_*

15 years agoRemove temporary debug tracing for curl_socklen_t detection failures
Yang Tse [Tue, 28 Apr 2009 09:40:12 +0000 (09:40 +0000)]
Remove temporary debug tracing for curl_socklen_t detection failures

15 years agoInclude <arpa/inet.h> if HAVE_ARPA_INET_H is defined
Yang Tse [Mon, 27 Apr 2009 23:59:41 +0000 (23:59 +0000)]
Include <arpa/inet.h> if HAVE_ARPA_INET_H is defined

15 years agoLog MAKEFLAGS environment variable along with the others.
Yang Tse [Mon, 27 Apr 2009 19:21:45 +0000 (19:21 +0000)]
Log MAKEFLAGS environment variable along with the others.

15 years agoTake 2 at handling getpeername() prototypes with a void pointer for third argument
Yang Tse [Mon, 27 Apr 2009 14:08:24 +0000 (14:08 +0000)]
Take 2 at handling getpeername() prototypes with a void pointer for third argument

15 years agoAttempt to handle getpeername() prototypes with a void pointer for third argument
Yang Tse [Mon, 27 Apr 2009 12:41:05 +0000 (12:41 +0000)]
Attempt to handle getpeername() prototypes with a void pointer for third argument

15 years agoAdd missing cmake files to the tarball (thanks to Richard Atterer's report)
Daniel Stenberg [Mon, 27 Apr 2009 12:26:41 +0000 (12:26 +0000)]
Add missing cmake files to the tarball (thanks to Richard Atterer's report)

15 years agothe windows threaded resolver isn't foolproof enough
Daniel Stenberg [Mon, 27 Apr 2009 12:23:53 +0000 (12:23 +0000)]
the windows threaded resolver isn't foolproof enough

15 years agoAdd temporary debug tracing for curl_socklen_t detection failures
Yang Tse [Mon, 27 Apr 2009 10:13:23 +0000 (10:13 +0000)]
Add temporary debug tracing for curl_socklen_t detection failures

15 years agoReorder curl_socklen_t and socklen_t equivalent check while both coexist
Yang Tse [Mon, 27 Apr 2009 02:47:34 +0000 (02:47 +0000)]
Reorder curl_socklen_t and socklen_t equivalent check while both coexist

15 years agoRearrange curl_socklen_t tests to improve speed of usual results
Yang Tse [Mon, 27 Apr 2009 00:17:11 +0000 (00:17 +0000)]
Rearrange curl_socklen_t tests to improve speed of usual results

15 years agoInitial step towards a configure time curl_socklen_t definition
Yang Tse [Sun, 26 Apr 2009 18:51:03 +0000 (18:51 +0000)]
Initial step towards a configure time curl_socklen_t definition

15 years agoFixed: 227 - CURLINFO_APPCONNECT_TIME doesn't work with multi interface, #2779733
Daniel Stenberg [Sun, 26 Apr 2009 11:57:45 +0000 (11:57 +0000)]
Fixed: 227 - CURLINFO_APPCONNECT_TIME doesn't work with multi interface, #2779733

15 years ago- Bug report #2779733 (http://curl.haxx.se/bug/view.cgi?id=2779733) by Sven
Daniel Stenberg [Sun, 26 Apr 2009 11:56:22 +0000 (11:56 +0000)]
- Bug report #2779733 (curl.haxx.se/bug/view.cgi?id=2779733) by Sven
  Wegener pointed out that CURLINFO_APPCONNECT_TIME didn't work with the multi
  interface and provided a patch that fixed the problem!

15 years agoremoved pointless file
Daniel Stenberg [Sat, 25 Apr 2009 21:01:28 +0000 (21:01 +0000)]
removed pointless file

15 years agosynced with current reality
Daniel Stenberg [Sat, 25 Apr 2009 20:56:09 +0000 (20:56 +0000)]
synced with current reality

15 years agoFurther narrow the use of the icc 9.1 optimizer workaround.
Yang Tse [Sat, 25 Apr 2009 10:24:11 +0000 (10:24 +0000)]
Further narrow the use of the icc 9.1 optimizer workaround.

Previous workaround proved useful, and finally did not trigger any warning!