|
From: Julian S. <js...@ac...> - 2011-10-25 10:32:23
|
A couple of hours ago I created branches/VALGRIND_3_7_BRANCH and branches/VEX_3_7_BRANCH, from which the 3.7.0 release will be made. It is not yet in a releaseable state: the documentation needs updating, the stuff in docs/internals/release-HOWTO.txt needs to get done, and it needs to be tested on all the supported platforms. There are also some last minute Android fixes to add. If all goes well I hope to have a release candidate tarball next Monday. Any testing that you can do prior to that point would be appreciated. I will keep track of, and merge, any changes on trunk that should also go in the branch, and vice versa. If you have a change that you think should get merged, please let me know. J |
|
From: Dave G. <go...@mc...> - 2011-10-25 18:52:37
Attachments:
vg_m4_quoting_r12230.patch
|
On Oct 25, 2011, at 5:29 AM CDT, Julian Seward wrote: > A couple of hours ago I created branches/VALGRIND_3_7_BRANCH and > branches/VEX_3_7_BRANCH, from which the 3.7.0 release will be made. > It is not yet in a releaseable state: the documentation needs > updating, the stuff in docs/internals/release-HOWTO.txt needs to get > done, and it needs to be tested on all the supported platforms. There > are also some last minute Android fixes to add. Does this mean that the patch attached to this bugzilla ticket will not make it into 3.7.0? Including it would make my life a bit easier and eliminate a common source of confusing build failures for users. 274078 improved configure logic for mpicc https://bugs.kde.org/show_bug.cgi?id=274078 I just attached a rebased version to the ticket and retested it on Darwin 10.6 and 64-bit Linux. Also, some m4 quoting is still missing after the recent build system updates. Autoconf 2.68 warns about this at autogen.sh-time. I've attached a small patch to this email to fix it. Thanks, -Dave |
|
From: John R. <jr...@bi...> - 2011-10-26 19:38:12
|
On 10/25/2011 03:29 AM, Julian Seward wrote: > > A couple of hours ago I created branches/VALGRIND_3_7_BRANCH and > branches/VEX_3_7_BRANCH, from which the 3.7.0 release will be made. [snip] I see complaints from autogen.sh that I have not seen before: ----- $ svn co svn://svn.valgrind.org/valgrind/branches/VALGRIND_3_7_BRANCH valgrind-3.7.branch A valgrind-3.7.branch/glibc-2.X.supp.in [snip many lines similar to above; appears to succeed] $ cd valgrind-3.7.branch $ ./autogen.sh running: aclocal configure.in:1344: warning: AC_LANG_CONFTEST: no AC_LANG_SOURCE call detected in body ../../lib/autoconf/lang.m4:194: AC_LANG_CONFTEST is expanded from... ../../lib/autoconf/general.m4:2591: _AC_COMPILE_IFELSE is expanded from... ../../lib/autoconf/general.m4:2607: AC_COMPILE_IFELSE is expanded from... ../../lib/m4sugar/m4sh.m4:606: AS_IF is expanded from... ../../lib/autoconf/general.m4:2591: _AC_COMPILE_IFELSE is expanded from... ../../lib/autoconf/general.m4:2607: AC_COMPILE_IFELSE is expanded from... configure.in:1344: the top level ----- and similar messages when the script continues with autoheader, automake -a, and [in VEX] autoconf. None of the messages appear when I run ./autogen.sh within a non-branched checkout. I'll try ignoring the messages on the branch, but am I missing something? -- |
|
From: Dave G. <go...@mc...> - 2011-10-26 20:43:40
|
On Oct 26, 2011, at 2:38 PM CDT, John Reiser wrote: > On 10/25/2011 03:29 AM, Julian Seward wrote: >> >> A couple of hours ago I created branches/VALGRIND_3_7_BRANCH and >> branches/VEX_3_7_BRANCH, from which the 3.7.0 release will be made. > [snip] > > I see complaints from autogen.sh that I have not seen before: > ----- > $ svn co svn://svn.valgrind.org/valgrind/branches/VALGRIND_3_7_BRANCH valgrind-3.7.branch > A valgrind-3.7.branch/glibc-2.X.supp.in > [snip many lines similar to above; appears to succeed] > $ cd valgrind-3.7.branch > $ ./autogen.sh > running: aclocal > configure.in:1344: warning: AC_LANG_CONFTEST: no AC_LANG_SOURCE call detected in body I saw similar messages on a recent trunk checkout. The patch attached to my previous mail on this thread fixes these warnings. One level of m4 quoting is missing from a small portion of configure.in, which autoconf-2.68 (and AFAIK only 2.68) detects and warns about. -Dave |
|
From: John R. <jr...@bi...> - 2011-10-26 21:05:48
|
>> running: aclocal >> configure.in:1344: warning: AC_LANG_CONFTEST: no AC_LANG_SOURCE call detected in body > > I saw similar messages on a recent trunk checkout. The patch attached to my previous mail on this thread fixes these warnings. One level of m4 quoting is missing from a small portion of configure.in, which autoconf-2.68 (and AFAIK only 2.68) detects and warns about. Thank you, Dave. vg_m4_quoting_r12230.patch works for me, too. -- |
|
From: Julian S. <js...@ac...> - 2011-10-26 21:20:16
|
On Wednesday, October 26, 2011, John Reiser wrote: > >> running: aclocal > >> configure.in:1344: warning: AC_LANG_CONFTEST: no AC_LANG_SOURCE call > >> detected in body > > > > I saw similar messages on a recent trunk checkout. The patch attached to > > my previous mail on this thread fixes these warnings. One level of m4 > > quoting is missing from a small portion of configure.in, which > > autoconf-2.68 (and AFAIK only 2.68) detects and warns about. > > Thank you, Dave. vg_m4_quoting_r12230.patch works for me, too. Ok, I'll push it into trunk and 3_7_BRANCH. Thanks for the patch and testing. Nothing if not arcane -- having looked at the patch makes me think of lisp, what with all those various kinds of brackets. J |
|
From: John R. <jr...@bi...> - 2011-10-27 17:42:16
|
On 10/25/2011 03:29 AM, Julian Seward wrote: > If you have a change that you think > should get merged, please let me know. https://bugs.kde.org/show_bug.cgi?id=276897#c7 contains a patch for faster dispatching and for better TT_FAST caching in ARM mode. It is derived from the earlier patch which also supports armv5/armv6. The new patch omits the changes associated with armv5/armv6 support, and so the resulting code still requires armv7 or above, just like the current VALGRIND_3_7_BRANCH. Thus the c7 patch aims to provide speed benefits without meddling in the issue of hardware support. Because I have only armv5te, then I cannot test the new patch. But if other developers do run the tests and the tests PASS, then I believe this is an appropriate candidate for merging. -- |