- developers-list - monetdb.org

Re: MonetDB: transaction-replication - Silence compiler
by Sjoerd Mullender 21 Jul '14

21 Jul '14

Not the right fix. __int64_t is not portable. In this case, since strings are never so long, a cast to (int) would have been good enough. You could also use ptrdiff_t which is the actual (standard C) type for the difference of pointers. On 2014-07-20 21:52, Dimitar Nedev wrote: > Changeset: 697c8e1dfbac for MonetDB > URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=697c8e1dfbac > Modified Files: > gdk/gdk_storage.c > Branch: transaction-replication > Log Message: > > Silence compiler > Avoid conversion from int64 to int > > > diffs (21 lines): > > diff --git a/gdk/gdk_storage.c b/gdk/gdk_storage.c > --- a/gdk/gdk_storage.c > +++ b/gdk/gdk_storage.c > @@ -100,7 +100,7 @@ GDKfilepath(int farmid, const char *dir, > char * > GDKfilepath_long(int farmid, const char *dir, const char *ext) { > char *last_dir_with_sep; > - int last_dirsep_index; > + __int64_t last_dirsep_index; > char *last_dir; > char *last_dir_parent; > > @@ -258,7 +258,7 @@ GDKfileopen(int farmid, const char * dir > /* if name is null, try to get one from dir (in case it was a path) */ > if ((name == NULL) || (*name == 0)) { > char *last_dir_with_sep; > - int last_dirsep_index; > + __int64_t last_dirsep_index; > char *last_dir; > char *last_dir_parent; > > _______________________________________________ > checkin-list mailing list > checkin-list(a)monetdb.org > https://www.monetdb.org/mailman/listinfo/checkin-list > -- Sjoerd Mullender

2 1

Announcement: New Jan2014-SP2 Bugfix release of MonetDB suite
by Sjoerd Mullender 02 Jun '14

02 Jun '14

The MonetDB team at CWI/MonetDB BV is pleased to announce the Jan2014-SP2 bugfix release of the MonetDB suite of programs. More information about MonetDB can be found on our website at <http://www.monetdb.org/>. For details on this release, please see the release notes at <http://www.monetdb.org/Downloads/ReleaseNotes>. As usual, the download location is <http://dev.monetdb.org/downloads/>. Jan 2014-SP2 bugfix release Build Environment * Lots of minor fixes were made for potential defects found by Coverity Scan. * On Windows we now build the geom module against version 3.4.2 of the geos library. Client Package * ODBC: Implemented {call procedure-name(...)} escape. The version {?=call ...} is not implemented. Bug Fixes * 3424: numeric values at the front of strings determines whether CAST works successfully * 3439: Python driver drops milliseconds from timestamps * 3446: SET READ ONLY forgets previous changes * 3455: String columns unusable from 64-bit .NET via ODBC * 3456: Insert fails * 3457: When kernel of remote client crashes, the connection remains established on server side * 3458: mserver5 crash on SQL: SELECT COUNT(*) FROM SYS.TABLES HAVING COUNT(*) > 0 * 3461: mserver5 crash on SQL: SELECT * FROM SYS.ARGS WHERE FUNC_ID NOT IN (SELECT ID FROM SYS.FUNCTIONS) OR FUNC_ID NOT IN (SELECT * FROM SYS.FUNCTIONS) * 3462: Invalid SQL (IN with subquery which returns multiple columns) is accepted * 3463: Crash on SELECT with SERIAL aggregation and GROUP BY column alias's * 3468: Local temporary table persists across sessions * 3469: Absolute network paths considered invalid for COPY INTO ... FROM statement. * 3473: Various memory leaks in SQL compilation * 3477: ODBC driver raises "unexpected end of input" for prepared string parameter from .NET application * 3481: Cannot run multiple COPY INTO statements in one 's'-command * 3482: Crossproduct error

2 1

TPC-H in MALs
by Qing Guo 21 Apr '14

21 Apr '14

Hi, I was trying to run “mserver” (standalone) in a simulator to do some performance study. As mserver only reads MALs, are there any ready-to-use MAL scripts for TPC-H? I tried to use “explain” in “mclient”, but the translated MALs seem to be too detailed and cannot simply copy and run. Any idea on this? Thank you! -Kevin

1 0

headless documentation
by Fabian Groffen 14 Mar '14

14 Mar '14

Hi devs, Is there any documentation for headless? Not that it seems completely pointless, but how exactly is it supposed to work? I'm looking for documentation that explains why, but preferably also old situation and how it should be converted into new situation kind of examples. Things like: can I simply insert into a void-headed BAT to automatically create a BAT with gaps if I e.g. insert 1, 5 and 7. Can I do a selection on the head (which should be a simple offset jump) to easily retrieve its tail, and how does the subselect interface exactly work. There appears to be not even comments on this piece of code. Regards from Fuerteventura, Fabian

2 1

Re: MonetDB: Jan2014 - configure.ag: revised way to determine compil...
by Fabian Groffen 11 Mar '14

11 Mar '14

On 05-03-2014 08:25:38 +0100, Stefan Manegold wrote: > Changeset: 36afc568f849 for MonetDB > URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=36afc568f849 > Modified Files: > configure.ag > Branch: Jan2014 > Log Message: > > configure.ag: revised way to determine compiler type (gcc vs. clang vs. icc): > in particular do not (only) rely on the result of AC_PROG_CC() (i.e., $GCC) > or the name of the call; > background is that on Mac OSX Mavericks, Apple prestends that their clang > were gcc ... FYI: they were pretending they had a GCC since OSX Lion. Fabian

4 14

Announcement: New Jan2014-SP1 Bugfix release of MonetDB suite
by Sjoerd Mullender 10 Mar '14

10 Mar '14

The MonetDB team at CWI/MonetDB BV is pleased to announce the Jan2014-SP1 bugfix release of the MonetDB suite of programs. More information about MonetDB can be found on our website at <http://www.monetdb.org/>. For details on this release, please see the release notes at <http://www.monetdb.org/Downloads/ReleaseNotes>. As usual, the download location is <http://dev.monetdb.org/downloads/>. Jan 2014-SP1 bugfix release Build Environment * Configure now enables the SQL front end by default. Merovingian * monetdb destroy -f now also works on running databases Bug Fixes * 3442: COPY INTO ... LOCKED reports incorrect count * 3443: DROP INDEX crashes server with BATsubselect: invalid argument: b must have a dense head * 3444: AND after ON () of LEFT OUTER JOIN with certain expressions will cause crash * 3452: ODBC driver build fails on Mac OS X due to a conflicting types for the SQLColAttribute with the unixODBC library

1 0

Re: MonetDB: Jan2014 - configure: don't choke on deprecation messages
by Stefan Manegold 06 Mar '14

06 Mar '14

Hi Fabian, thank you very much for your fix! Do I understand correctly that the problem is in fact related to (recent versions of) MacOSX, and the packages / libraries that come with it, rather than the compiler itself? If so, we might want to consider adding option "-Wno-deprecated-declarations" only on (the respective versions of) MacOSX, --- or only when there is Apple's CommonCrypto API rather than OpenSSL, provided we can easily detect that --- rather than whenever using any version of clang on any system (and with old versions of gcc on any system). Yes, I still have the ambition to disable strict compilation checks only as selectively as possible, even on "odd" platforms ... ;-) Thanks! Stefan ----- Original Message ----- > Changeset: b4d844547d39 for MonetDB > URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=b4d844547d39 > Modified Files: > configure.ag > Branch: Jan2014 > Log Message: > > configure: don't choke on deprecation messages > > Apple "deprecated" OpenSSL since Lion, don't complain about that. > > > diffs (15 lines): > > diff --git a/configure.ag b/configure.ag > --- a/configure.ag > +++ b/configure.ag > @@ -598,8 +598,11 @@ yes-*-*) > dnl not all) false positives, though, as well as in > dnl YACC/BISON-generated code; thus, we do not use > dnl "-Wunreachable-code" with clang and gcc < 4.5 . > + dnl At the same time, Apple deprecated OpenSSL in favour of > + dnl their own CommonCrypto API. We don't mind about that. > case "$CC_ver" in > clang-*|gcc-[[0-3]].*|gcc-4.[[0-4]].*) > + MCHECK_ADD_FLAG([-Wno-deprecated-declarations]) > ;; > *) > MCHECK_ADD_FLAG([-Wunreachable-code]) > _______________________________________________ > checkin-list mailing list > checkin-list(a)monetdb.org > https://www.monetdb.org/mailman/listinfo/checkin-list > -- | Stefan.Manegold(a)CWI.nl | DB Architectures (DA) | | www.CWI.nl/~manegold/ | Science Park 123 (L321) | | +31 (0)20 592-4212 | 1098 XG Amsterdam (NL) |

2 1

Re: MonetDB: Jan2014 - configure.ag: with clang, disable -pthread as...
by Fabian Groffen 05 Mar '14

05 Mar '14

On 05-03-2014 08:25:38 +0100, Stefan Manegold wrote: > Changeset: a06afa30eeea for MonetDB > URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=a06afa30eeea > Modified Files: > configure.ag > Branch: Jan2014 > Log Message: > > configure.ag: with clang, disable -pthread as compiler option > (but keep it as linker option) > > Also clang 3.3 on fedora 20 appears to work fine without > -pthread as compiler option (but requires it as linker option). from man gcc: -pthread Adds support for multithreading with the pthreads library. This option sets flags for both the preprocessor and linker. This is typically something like _REENTRANT. On Mountain Lion, gcc/clang would set _REENTRANT. Same for CentOS 6. % /usr/bin/gcc -dumpspecs | grep pthread %{static:%{!dynamic:-D__STATIC__}}%{!static:-D__DYNAMIC__} %%{pthread:-D_REENTRANT} What I can imagine here, is that because clang on Mavericks now defaults to C99 mode (instead of gnu89 mode gcc defaults to) you get different defines and behaviour, possibly including _REENTRANT. A preprocessor check would have to give some facts around that. Fabian

1 0

status of default branch
by Fabian Groffen 05 Mar '14

05 Mar '14

Hi all, Since I'm no longer really in the loop with MonetDB, I was wondering what the state of the default branch is supposed to be. I just fixed a configure error and now hit a redeclaration problem regarding gdk.h and gdk_private.h. gdk_batop.c includes both headers so it seems like a valid complaint to me. Is OSX still a supported platform, these days? I guess there is little to no point in supporting it. Fabian

5 20

Building the ODBC driver on OS X 10.9.2
by Dimitar Nedev 04 Mar '14

04 Mar '14

Hello everybody, I managed to build the latest MonetDB version on my Mac with all default modules, excluding the DataCell and RDF, but the ODBC build fails. Trying to fix it I encountered several issues: A few libraries could not be found with the default build settings: hwcounters = no (no supported harwdware counters interface found) mseed = no (libmseed.h header not found) raptor = no (by default) readline = no (readline/readline.h does not contain rl_completion_func_t, is it GNU readline?) setsockopt = no (by default) sphinxclient = no (sphinxclient.h header not found) unixodbc = no (odbcinst.h not found) valgrind = no (by default) So far it makes sense, since unixODBC cannot be found, the ODBC driver cannot be built. Just that I've actually installed the library. Adding the following line: --with-unixodbc=/opt/local --with-sphinxclient=/opt/local --with-readline=/opt/local to point to some of the libs location appeared to fix the configuration: checking odbcinst.h usability... yes checking odbcinst.h presence... yes checking for odbcinst.h... yes checking for SQLGetPrivateProfileString in -lodbcinst... yes checking for SQLGetDiagRec in -lodbc... yes checking checking whether SQLColAttribute uses SQLLEN * or SQLPOINTER... using "SQLPOINTER" hwcounters = no (no supported harwdware counters interface found) mseed = no (libmseed.h header not found) raptor = no (by default) setsockopt = no (by default) valgrind = no (by default) Now, trying to build that configuration yields the following error: libtool: compile: gcc -DHAVE_CONFIG_H -I. -I../../../../clients/odbc/driver -I../../.. -I../../../../clients/odbc/driver -I../../mapilib -I../../../../clients/odbc/driver/../../mapilib -I../../../common/options -I../../../../clients/odbc/driver/../../../common/options -I/opt/local/include -DLIBMONETODBC -g -O2 -I/opt/local/include -c ../../../../clients/odbc/driver/SQLColAttributes.c -fno-common -DPIC -o .libs/libMonetODBC_la-SQLColAttributes.o ../../../../clients/odbc/driver/SQLColAttribute.c:243:1: error: conflicting types for 'SQLColAttribute' SQLColAttribute(SQLHSTMT StatementHandle, ^ /opt/local/include/sql.h:609:24: note: previous declaration is here SQLRETURN SQL_API SQLColAttribute(SQLHSTMT StatementHandle, ^ ../../../../clients/odbc/driver/SQLColAttribute.c:272:1: error: conflicting types for 'SQLColAttributeA' SQLColAttributeA(SQLHSTMT StatementHandle, ^ /opt/local/include/sqlucode.h:377:19: note: previous declaration is here SQLRETURN SQL_API SQLColAttributeA( ^ ../../../../clients/odbc/driver/SQLColAttribute.c:290:1: error: conflicting types for 'SQLColAttributeW' SQLColAttributeW(SQLHSTMT StatementHandle, ^ /opt/local/include/sqlucode.h:29:19: note: previous declaration is here SQLRETURN SQL_API SQLColAttributeW( ^ 3 errors generated. make[7]: *** [libMonetODBC_la-SQLColAttribute.lo] Error 1 I am running: OS X 10.9.2 / Darwin Kernel Version 13.1.0 x86_64 Xcode 5.0.2 and gcc -v reports: Apple LLVM version 5.0 (clang-500.2.79) (based on LLVM 3.3svn) unixODBC 2.3.1_0 installed from MacPorts 2.2.1 Any ideas? Thanks, Dimitar

2 2