Jack Howarth | 19 Apr 22:49 2015
Picon

boost1.55.info missing libicu depends

Hanspeter,
       I stumbled across a packaging flaw in boost1.53..info and
boost1.55.info while attempting to migrate the studio-desktop package
to boost 1.55. The package built fine but crashed at runtime with the
error...

Dyld Error Message:
  Library not loaded: /sw/lib/libicuuc.48.1.1.dylib
  Referenced from: /sw/opt/boost-1_55/lib/libboost_regex-mt-1_55.dylib
  Reason: image not found

It appears that we need a BuildDepends and Depends line for the libicu
that is linked into libboost_regex-mt-1_55.dylib...

# otool -L /sw/opt/boost-1_55/lib/libboost_regex-mt-1_55.dylib
/sw/opt/boost-1_55/lib/libboost_regex-mt-1_55.dylib:
/sw/opt/boost-1_55/lib/libboost_regex-mt-1_55.dylib (compatibility
version 1.55.0, current version 1.55.0)
/sw/lib/libicuuc.48.1.1.dylib (compatibility version 48.0.0, current
version 48.1.1)
/sw/lib/libicui18n.48.1.1.dylib (compatibility version 48.0.0, current
version 48.1.1)
/sw/lib/libicudata.48.1.1.dylib (compatibility version 48.0.0, current
version 48.1.1)
/usr/lib/libc++.1.dylib (compatibility version 1.0.0, current version 120.0.0)
/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current
version 1213.0.0)

We need to make the change to both boost1.53..info and boost1.55.info

(Continue reading)

Jack Howarth | 18 Apr 22:06 2015
Picon

Re: rstudio-desktop-0.98.1103-1

I've also added a rstudio-server.info into my experimental which uses
the same build approach to package 0.98.1103 against libc++ on
10.9/10.10.

On Sat, Apr 18, 2015 at 1:18 PM, Jack Howarth <howarth.at.fink <at> gmail.com> wrote:
> FYI, I posted the two libc++ compilation errors to the Rstudio support
> pages in case they have any uncommitted fixes upstream.
>
> https://support.rstudio.com/hc/communities/public/questions/203373368-Rstudio-0-98-1103-has-only-two-libc-build-issues-left?locale=en-us
>
> I also posted a message on the duplicate symbol issue introduce when
> building against R 3.2.0 and the workaround I am using.
>
> https://support.rstudio.com/hc/communities/public/questions/203671967-duplicate-R-running-as-main-program-break-linkage-of-rstudio-desktop-0-98-1103-against-R-3-2-0?locale=en-us
>
>
>
> On Sat, Apr 18, 2015 at 12:03 PM, Jack Howarth
> <howarth.at.fink <at> gmail.com> wrote:
>> Baba,
>>      I've updated the rstudio-desktop.info file in my experimental to
>> use a fixincludes subdirectory to provide the build with a local copy
>> of the 3.2 Rinterface.h header with the extern restored to
>> R_running_as_main_program, This eliminated the duplicate symbols for
>> R_running_as_main_program in the final linkage when built against
>> r-base32 and I've switched that package from 3.1 to 3.2. Now we only
>> have the two libc++ compilation bugs in
>> src/cpp/core/r_util/RSessionLaunchProfile.cpp  and
>> src/cpp/session/modules/SessionHTMLPreview.cpp to properly solve.
>>            Jack
(Continue reading)

Jack Howarth | 18 Apr 17:33 2015
Picon

rstudio-desktop-0.98.1103-1

Baba,
       As you probably noticed, I added r-base32-3.2.0-1 packaging to
fink this week. I also updated all the R modules packages in
libs/rmods to support 3.2 as well as updating them to their most
current upstream cran versions.
      With that done, I've resurrected the rstudio-desktop package by
updating it to the current 0.98.1103 release and tweaking the build to
complete on 10.9/10.10. The changes are roughly...

1) perl patching src/cpp/CMakeLists.txt on darwin13 or later to pass
-stdlib=libc++ rather than -stdlib=stdlibc++.
2) apply an ugly hack in rstudio-desktop-libc++.patch  to disable two
compile errors against libc++ until we craft a proper fix.

I have left the build on boost 1.53. While the build against boost
1.55 completes, the resulting application crashes on startup. Also,
rstudio-desktop is currently stuck on r-base31 as a change in 3.2.0
breaks the executable linkage due to multiple symbols for
_R_running_as_main_program.

http://r.789695.n4.nabble.com/Removed-extern-on-R-running-as-main-program-in-Rinterface-h-intentional-tt4694398.html#none

Perhaps we can kludge around that by making a local copy of the
offending R header without the declaration and manually add the pruned
lines back into only a single rstudio file.
     You can find the packaging files...

rstudio-desktop.info
rstudio-desktop.patch
rstudio-desktop-libc++.patch
(Continue reading)

Jack Howarth | 17 Apr 21:26 2015
Picon

debug strigi packaging for 10.9

Hanspeter,
      You can use the attached packaging with the following changes to
debug the strigi build issue on 10.9 against Xcode 6.2 by substituting
the clang-3.5/clang++-3.5 compilers from llvm35-3.5.2 which also
exhibits the new strictness.
                 Jack

--- strigi.info 2015-04-17 15:20:35.000000000 -0400
+++ strigi.info.debug 2015-04-17 15:18:13.000000000 -0400
 <at>  <at>  -18,6 +18,8  <at>  <at> 
 <<
 BuildDepends: <<
  bzip2-dev,
+ clang35,
+ libcxx1-dev,
  clucene-core (>= 0.9.20-1),
  cmake (>= 2.8.10.2-1),
  cppunit1.12.1,
 <at>  <at>  -28,8 +30,12  <at>  <at> 
  libxml2 (>= 2.9.1-1),
  pkgconfig (>= 0.23-1)
 <<
+BuildConflicts: strigi-dev
 GCC: 4.0

+SetCC: clang-3.5
+SetCXX: clang++-3.5
+
 Source: http://www.vandenoever.info/software/strigi/strigi-%v.tar.bz2
 Source-MD5: d69443234f4286d71997db9de543331a
(Continue reading)

Jack Howarth | 16 Apr 20:47 2015
Picon

storage build broken under Xcode 6.3

The Apple Clang 6.1 compilers from Xcode 6.3 break the build of the
strigi package with the error...

[ 92%] Building CXX object
strigidaemon/bin/daemon/xesam/CMakeFiles/xesam.dir/xesamsession.cpp.o
cd /sw/src/fink.build/strigi-0.7.8-2/strigi-0.7.8/build/strigidaemon/bin/daemon/xesam
&& /sw/var/lib/fink/path-prefix-clang/c++    -MD -isysroot
/Applications/Xcode.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX10.10.sdk
-mmacosx-version-min=10.10
-I/sw/src/fink.build/strigi-0.7.8-2/strigi-0.7.8/libstreams/include
-I/sw/src/fink.build/strigi-0.7.8-2/strigi-0.7.8/build/libstreams/include
-I/sw/src/fink.build/strigi-0.7.8-2/strigi-0.7.8/libstreamanalyzer/include
-I/sw/src/fink.build/strigi-0.7.8-2/strigi-0.7.8/build/libstreamanalyzer/include
-I/sw/src/fink.build/strigi-0.7.8-2/strigi-0.7.8/strigidaemon/include
-I/sw/src/fink.build/strigi-0.7.8-2/strigi-0.7.8/strigidaemon/bin/daemon/xsd
-I/sw/src/fink.build/strigi-0.7.8-2/strigi-0.7.8/build/strigidaemon
-I/sw/src/fink.build/strigi-0.7.8-2/strigi-0.7.8/build/strigidaemon/bin/daemon/dbus
-I/sw/include/dbus-1.0 -I/sw/lib/dbus-1.0/include -I/sw/include
-I/sw/include/libxml2
-I/sw/src/fink.build/strigi-0.7.8-2/strigi-0.7.8/strigidaemon/bin/combinedindex
-I/sw/src/fink.build/strigi-0.7.8-2/strigi-0.7.8/strigidaemon/bin/daemon
   -fPIC -o CMakeFiles/xesam.dir/xesamsession.cpp.o -c
/sw/src/fink.build/strigi-0.7.8-2/strigi-0.7.8/strigidaemon/bin/daemon/xesam/xesamsession.cpp
In file included from
/sw/src/fink.build/strigi-0.7.8-2/strigi-0.7.8/strigidaemon/bin/daemon/xesam/xesamsession.cpp:22:
In file included from
/sw/src/fink.build/strigi-0.7.8-2/strigi-0.7.8/strigidaemon/bin/daemon/xesam/xesamsession.h:24:
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/../include/c++/v1/list:2060:18:
error: invalid operands to binary expression
      ('const value_type' (aka 'const XesamSearch') and 'const
(Continue reading)

Alexander Hansen | 16 Apr 18:46 2015
Picon

Handling flat_namespace on 10.10+

Summary:  GNU libtool effectively assumed that there would be no 10.10, so a bunch of packages inherited
conditional logic that treats 10.10 like 10.1.  We’ve been patching against this, and put a .deb
validator check for flat_namespace builds.

Problem:  openmpi apparently requires flat_namespace.  Other packages might also need it, too, but I
don’t happen to know of any offhand.

There are a couple of options to address the problem:

1)  Add a boolean override field, e.g. BuildFlatNamespace, to the .info and have that turn off the .deb
validation check.
2)  Get rid of the .deb validation check and instead apply mandatory tests in the earlier phases.  For
example, to test at the end of the compile phase fink-package-precedence could be extended also to check
for flat_namespace and packages which need flat_namespace wouldn’t use f-p-p; or an additional
option flag could be added to f-p-p.  We could also check config.status before the compile phase.

The virtue of option 1 is that by the time the .debs are generated most of the weirdness in particular build
systems has been smoothed over, so that we’re working with libraries and executables.  The downside is
that the whole package gets built before a problem is found.

Option 2 involves updating f-p-p and generating other helper scripts, or perhaps have a check in fink and
use a boolean override flag to turn that check off—that way the default CompileScript would include the test.

Thoughts?

--

-- 
Alexander Hansen, Ph.D.
Fink User Liaison

------------------------------------------------------------------------------
(Continue reading)

Jack Howarth | 16 Apr 14:39 2015
Picon

mysql-unified-5.0.96-4 validation failures

   The current mysql-unified-5.0.96-4 fails to validate with...

Stopping All Servers
Failed 7/477 tests, 98.53% were successful.

The log files in var/log may give you some hint
of what went wrong.
If you want to report this error, please read first the documentation at
http://dev.mysql.com/doc/mysql/en/mysql-test-suite.html
The servers were restarted 128 times
Spent 440.119 of 2741 seconds executing testcases

mysql-test-run in default mode: *** Failing the test(s): openssl_1
rpl_openssl rpl_ssl ssl ssl_8k_key ssl_compress ssl_connect
mysql-test-run: *** ERROR: there were failing test cases
Makefile:929: recipe for target 'test-ns' failed
make[1]: *** [test-ns] Error 1
make[1]: Leaving directory
'/sw/src/fink.build/mysql-unified-5.0.96-4/mysql-5.0.96'
Makefile:938: recipe for target 'test-force' failed
make: *** [test-force] Error 2

Are these due to the broken daemonic or some other issues?
            Jack

------------------------------------------------------------------------------
BPM Camp - Free Virtual Workshop May 6th at 10am PDT/1PM EDT
Develop your own process in accordance with the BPMN 2 standard
Learn Process modeling best practices with Bonita BPM through live exercises
http://www.bonitasoft.com/be-part-of-it/events/bpm-camp-virtual- event?utm_
(Continue reading)

Jack Howarth | 12 Apr 22:45 2015
Picon

texlive_0.20140523-2

Tomoaki,
      The texlive package needs the following update committed...

Index: texlive.info
===================================================================
RCS file: /cvsroot/fink/dists/10.7/stable/main/finkinfo/text/texlive.info,v
retrieving revision 1.19
diff -r1.19 texlive.info
5c5
< Revision: 1
---
> Revision: 2
22c22
<  libicu48-dev,
---
>  libicu55-dev,
229c229
<   libicu48-shlibs,
---
>   libicu55-shlibs,

The Apple Clang 6.1 compilers from Xcode 6.3 builds a libicu48-dev
package with test suite regressions (and lowering the optimization to
-O0 doesn't eliminate those). I have added a libicu55-dev package
which builds without compiler warnings or test suite regressions under
Xcode 6.3. The attached texlive.info builds against this new
libicu55-dev/libicu55-shlibs and passes 'fink -m' on 10.10 with Xcode
6.3.
           Thanks in advance.
                 Jack
(Continue reading)

Jack Howarth | 12 Apr 21:55 2015
Picon

libicu48-dev broken by Xcode 6.3

    A head's up that the Apple clang 6.1 compiler in Xcode 6.3 causes
the libicu48-dev package to fail its InfoTest. There doesn't seem an
obvious fix as reducing the optimization down to -O0 fails to
eliminate the test suite regressions. Fortunately the icu 55.1 release
from last month builds without warnings and test suite failures
against Xcode 6.3. I have added a new libicu55-dev package and
migrated r-base31 and r-base32 over to it. I have left the InfoTest
failing in libicu48-dev for now to get more eyes on the issue but
suspect we will have to abandon the older releases on the newer clang
compilers (as it has a very noisy build with some ominous warnings of
array overruns).
             Jack

------------------------------------------------------------------------------
BPM Camp - Free Virtual Workshop May 6th at 10am PDT/1PM EDT
Develop your own process in accordance with the BPMN 2 standard
Learn Process modeling best practices with Bonita BPM through live exercises
http://www.bonitasoft.com/be-part-of-it/events/bpm-camp-virtual- event?utm_
source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF
_______________________________________________
Fink-devel mailing list
Fink-devel <at> lists.sourceforge.net
List archive:
http://news.gmane.org/gmane.os.apple.fink.devel
Subscription management:
https://lists.sourceforge.net/lists/listinfo/fink-devel

Jack Howarth | 11 Apr 14:48 2015
Picon

qhull6.3.1-2012.1-3 broken on Xcode 6.3's Apple Clang 6.1

The new Xcode 6.3's Apple Clang 6.1 compiler breaks the build of
qhull6.3.1-2012.1-3.

g++ -c -O2 -Dqh_QHpointer -Isrc/ -Isrc/libqhullcpp -Isrc/libqhull
-Wall -Wcast-qual -Wextra -Wwrite-strings  -Wshadow -Wconversion  -o
src/user_eg3/user_eg3.o src/user_eg3/user_eg3.cpp
...
src/libqhullcpp/Coordinates.h:154:22: error: reference to
'random_access_iterator_tag' is ambiguous
        typedef std::random_access_iterator_tag  iterator_category;
                     ^
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/../include/c++/v1/iterator:353:30:
note: candidate found by name lookup is
      'std::__1::random_access_iterator_tag'
struct _LIBCPP_TYPE_VIS_ONLY random_access_iterator_tag : public
bidirectional_iterator_tag {};
                             ^
src/libqhullcpp/QhullIterator.h:20:59: note: candidate found by name
lookup is 'std::random_access_iterator_tag'
namespace std { struct bidirectional_iterator_tag; struct
random_access_iterator_tag; }
                                                          ^
In file included from src/user_eg3/user_eg3.cpp:3:
In file included from src/libqhullcpp/RboxPoints.h:12:
In file included from src/libqhullcpp/QhullPoint.h:15:
src/libqhullcpp/Coordinates.h:201:22: error: reference to
'random_access_iterator_tag' is ambiguous
        typedef std::random_access_iterator_tag  iterator_category;
                     ^
/Applications/Xcode.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin/../include/c++/v1/iterator:353:30:
(Continue reading)

Jack Howarth | 10 Apr 20:54 2015
Picon

gcc49 bootstrap comparison failures

     Just a heads up that the Xcode 6.3 release currently won't build
the gcc49-4.9.2 package due to bootstrap comparison failures. The
issue is under investigation upstream and also occurs when the
clang-3.5/clang++-3.5 compilers from llvm-3.5.1/llvm-3.5.2 or the
clang-3.6/clang++-3.6 compilers from llvm-3.6.0 are used under Xcode
6.2. The latter observation should eliminate any changes in the
cctools of Xcode 6.3 as the cause and seems to firmly point to
miscompilation of the stage1 compilers by the newer clang releases.
          Jack
ps Oddly the upcoming gcc 5.0 compiler doesn't suffer the same
bootstrap comparison failure (but it is a pretty massive rewrite from
4.9.x). If we are still stuck without a fix in a week or so, I can
push out the gcc5 packaging from the most current snapshot (as it is
very close to the RC release).

------------------------------------------------------------------------------
BPM Camp - Free Virtual Workshop May 6th at 10am PDT/1PM EDT
Develop your own process in accordance with the BPMN 2 standard
Learn Process modeling best practices with Bonita BPM through live exercises
http://www.bonitasoft.com/be-part-of-it/events/bpm-camp-virtual- event?utm_
source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF
_______________________________________________
Fink-devel mailing list
Fink-devel <at> lists.sourceforge.net
List archive:
http://news.gmane.org/gmane.os.apple.fink.devel
Subscription management:
https://lists.sourceforge.net/lists/listinfo/fink-devel

(Continue reading)


Gmane