- developers-list - monetdb.org

[Monetdb-developers] [PATCH] Loop Lifted Pathfinder sources
by Fabian Groffen 03 Nov '06

03 Nov '06

Hi, Please find attached a (compressed) patch which loop lifts all loops in the pathfinder sources such that variable declarations go at the top of the scope the loop is in, or a higher scope if more appropriate. Rationale: I don't feel like appending stdc99 flags in a (possible) Gentoo ebuild for the next release. In other words, this patch removes the C99 requirement, which I think more people will benefit from, such that MonetDB/* cleanly compiles with --disable-strict. Please consider applying this patch to the current CVS sources. My apologies for the large attachment on this list; there are simply lots of loops to "lift". -- Fabian Groffen Gentoo on a different level

7 12

Re: [Monetdb-developers] [Monetdb-sql-checkins] sql/src/backends/monet5 sql.mx, 1.190, 1.191 sql_scenario.mx, 1.221, 1.222
by Sjoerd Mullender 03 Nov '06

03 Nov '06

On 2006-11-03 10:02, Niels Nes wrote: > Update of /cvsroot/monetdb/sql/src/backends/monet5 > In directory sc8-pr-cvs7.sourceforge.net:/tmp/cvs-serv20243/src/backends/monet5 > > Modified Files: > sql.mx sql_scenario.mx > Log Message: > sql_scenario: reset stream by seting eof=1 > > approved tests need because of logger changes > > Added test for identifier bug (ADT2006/sander.sql) > We now disallow invalied identifiers in double quotes (and don't > fallback to strings in those cases) > Approved tests affected by this change > > rel_optimize: icc compilation fix > > > Index: sql_scenario.mx > =================================================================== > RCS file: /cvsroot/monetdb/sql/src/backends/monet5/sql_scenario.mx,v > retrieving revision 1.221 > retrieving revision 1.222 > diff -u -d -r1.221 -r1.222 > --- sql_scenario.mx 27 Oct 2006 13:37:44 -0000 1.221 > +++ sql_scenario.mx 3 Nov 2006 09:02:54 -0000 1.222 > @@ -315,8 +315,10 @@ > /* TODO reset global variables */ > } > > - c->fdin->pos += c->yycur; > + //c->fdin->pos += c->yycur; Niels, you should know better than this... > c->yycur = 0; /* make sure we don't reuse the used input */ > + c->fdin->eof = 1; /* force prompt on first iteration */ > + > initSQLreferences(); > > return MAL_SUCCEED; > > Index: sql.mx > =================================================================== > RCS file: /cvsroot/monetdb/sql/src/backends/monet5/sql.mx,v > retrieving revision 1.190 > retrieving revision 1.191 > diff -u -d -r1.190 -r1.191 > --- sql.mx 27 Oct 2006 18:23:50 -0000 1.190 > +++ sql.mx 3 Nov 2006 09:02:54 -0000 1.191 > @@ -723,7 +723,6 @@ > b->currColumn = NULL; > b->currKey = NULL; > b->currIndex = NULL; > - //b->code = NULL; > b->vtop = 0; > b->Tparse = b->Toptimize = b->Tschedule = b->Texec = 0; > b->qry = NULL; > > > ------------------------------------------------------------------------- > Using Tomcat but need to do more? Need to support web services, security? > Get stuff done quickly with pre-integrated technology to make your job easier > Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 > _______________________________________________ > Monetdb-sql-checkins mailing list > Monetdb-sql-checkins(a)lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/monetdb-sql-checkins -- Sjoerd Mullender

2 1

Re: [Monetdb-developers] [Monetdb-pf-checkins] pathfinder/compiler/include compile_interface.h, 1.15, 1.16
by Stefan Manegold 03 Nov '06

03 Nov '06

On Fri, Nov 03, 2006 at 03:43:07AM +0000, Peter Boncz wrote: > Update of /cvsroot/monetdb/pathfinder/compiler/include > In directory sc8-pr-cvs7.sourceforge.net:/tmp/cvs-serv18716/compiler/include > > Modified Files: > compile_interface.h > Log Message: > new features: > - automatic serialized transaction restart and convoys for conflict-aborted > transactions. > - better garbage collection after delete_doc (let ws_destroy that blocked > destruction of a collection throw it away -- the zombie bat). > - proper support for updating functions > > some bugfixes: > - log changes in the qn_histogram column as well (it needs to be in sync) > - fixed some bugs in the coll_lock locking in pf_support > READS in the master bats are FORBIDDEN as well, unless the lock is held > > still lots of BUGS seen: > - (shred, insert, additive-shred) gives assert > - BBPname errors sfter restart > - with profiked modules, ws_create hangs on pfenter (?) > - nid_rid batgets is way too large (each session adds a page to id -- > maybe related to shredder assert) > - collection_cleanup should remove bats from deleted collection > from the logger administration please add your respective tests to CVS --- running them with standard testing (`make check` or `Mtest` before checkins, automatic nightly testing) will (a) remind us of open bugs/problems and (b) help us to ensure that the bugs stay fixed once they have been fixed. Thanks! Stefan -- | Dr. Stefan Manegold | mailto:Stefan.Manegold@cwi.nl | | CWI, P.O.Box 94079 | http://www.cwi.nl/~manegold/ | | 1090 GB Amsterdam | Tel.: +31 (20) 592-4212 | | The Netherlands | Fax : +31 (20) 592-4312 |

1 0

Re: [Monetdb-developers] [Monetdb-checkins] MonetDB/src/mapi/clients/php/Cimpl php_monetdb.c, 1.54, 1.55 php_monetdb.h, 1.20, 1.21
by Stefan Manegold 02 Nov '06

02 Nov '06

Fabian, thanks for the fixes! unfortunately, there are (at least I get) still quite some warnings with both gcc & icc (see attachments); hence, it's still too early to re-enable strict flags for php ... Stefan On Thu, Nov 02, 2006 at 02:15:00PM +0000, Fabian wrote: > Update of /cvsroot/monetdb/MonetDB/src/mapi/clients/php/Cimpl > In directory sc8-pr-cvs7.sourceforge.net:/tmp/cvs-serv4902 > > Modified Files: > php_monetdb.c php_monetdb.h > Log Message: > Resolve two warnings in our php module. It now compiles without any > warnings. I'm not quite sure how to enable strict flags for php again. > > > Index: php_monetdb.h > =================================================================== > RCS file: /cvsroot/monetdb/MonetDB/src/mapi/clients/php/Cimpl/php_monetdb.h,v > retrieving revision 1.20 > retrieving revision 1.21 > diff -u -d -r1.20 -r1.21 > --- php_monetdb.h 26 May 2006 08:45:35 -0000 1.20 > +++ php_monetdb.h 2 Nov 2006 14:14:58 -0000 1.21 > @@ -124,6 +124,10 @@ > #define MONETDB_DML_ASYNC (1<<10) /* Do async query */ > #define MONETDB_DML_STRING (1<<11) /* Return query string */ > > +/* avoid redeclaration */ > +#ifdef _POSIX_C_SOURCE > +# undef _POSIX_C_SOURCE > +#endif > > #include "Mapi.h" > /* typedefs for convenience; not using Mapi's as those already are a > > Index: php_monetdb.c > =================================================================== > RCS file: /cvsroot/monetdb/MonetDB/src/mapi/clients/php/Cimpl/php_monetdb.c,v > retrieving revision 1.54 > retrieving revision 1.55 > diff -u -d -r1.54 -r1.55 > --- php_monetdb.c 24 Oct 2006 14:26:18 -0000 1.54 > +++ php_monetdb.c 2 Nov 2006 14:14:58 -0000 1.55 > @@ -616,13 +616,13 @@ > */ > if (!(connect_type & MONETDB_CONNECT_FORCE_NEW) > && zend_hash_find(&EG(regular_list),str.c,str.len+1,(void **) &index_ptr)==SUCCESS) { > - int type,link; > + int type, link; > void *ptr; > > if (Z_TYPE_P(index_ptr) != le_index_ptr) { > RETURN_FALSE; > } > - link = (int) index_ptr->ptr; > + link = (int)(size_t) index_ptr->ptr; > ptr = zend_list_find(link,&type); /* check if the link is still there */ > if (ptr && (type==le_link || type==le_plink)) { > Z_LVAL_P(return_value) = link; > > > ------------------------------------------------------------------------- > Using Tomcat but need to do more? Need to support web services, security? > Get stuff done quickly with pre-integrated technology to make your job easier > Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 > _______________________________________________ > Monetdb-checkins mailing list > Monetdb-checkins(a)lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/monetdb-checkins > > > > -- | Dr. Stefan Manegold | mailto:Stefan.Manegold@cwi.nl | | CWI, P.O.Box 94079 | http://www.cwi.nl/~manegold/ | | 1090 GB Amsterdam | Tel.: +31 (20) 592-4212 | | The Netherlands | Fax : +31 (20) 592-4312 |

2 1

[Monetdb-developers] XQuery: cant reproduce testweb results
by Wouter Alink 01 Nov '06

01 Nov '06

Hello, I've been trying to compile and run MonetDB/XQuery (the head) for a few evenings now, but I can't get the results like the testweb shows. I'm using isoda.ins.cwi.nl, and was expecting the testweb to become as-green-as the FC4-64bits-testweb. - the last sources I compiled were from this evening (0.13.1) - During configure or make, their were no errors, neither in buildtools, monetdb or pathfinder. - Mtest_MonetDB shows almost completely green testresults (like the testWeb) - with Mtest_pathfinder almost half of the queries fail. Many of the failing tests show stdout output as provided below. I have >1.5 GB of free disk-space. By now I must have thrown away any part of the MonetDB system (including dbfarms and Monetdb.confs) at least twice, before trying again. Also, none of my environment variables seem to point to old installations of MonetDB. Does anybody have any clue what i'm doing wrong? All your help is welcome, Thanks in advance, Wouter p.s. Stefan: I noticed that Mtest.py sets sql_logdir twice. I guessed that at the second occasion it should set xquery_logdir. (lineno. 2586 in Mtest.py) ----- stdout of test 'ID.1187994-a` in directory 'tests/BugsViaSourgeforce` itself: # 22:36:07 > # 22:36:07 > Mtimeout -timeout 180 Mserver "--config=/ufs/alink/local/etc/MonetDB.conf" --debug=10 --set "monet_mod_path=/export/scratch1/alink/src/pathfinder-head/Linux/lib64/Monet DB:/export/scratch1/alink/src/pathfinder-head/Linux/lib64/MonetDB/lib:/ufs/a link/local/lib64/MonetDB:/ufs/alink/local/lib64/MonetDB/lib:/ufs/alink/local /lib64/MonetDB/bin" --set "gdk_dbfarm=/ufs/alink/local/var/MonetDB/dbfarm" --set "sql_logdir=/ufs/alink/local/var/MonetDB/sql_logs" --set "xquery_logdir=/ufs/alink/local/var/MonetDB/xquery_logs" --set mapi_port=30644 --set monet_prompt= --trace "--dbname=mTests_tests_BugsViaSourgeforce" --dbinit="module(pathfinder); pfstart();" ; echo ; echo Over.. # 22:36:07 > # Monet Database Server V4.13.1 # Copyright (c) 1993-2006, CWI. All rights reserved. # Compiled for x86_64-redhat-linux-gnu/64bit with 64bit OIDs; dynamically linked. # Visit http://monetdb.cwi.nl/ for further information. !ERROR: GDKload: cannot read: name=43, ext=theap, 40 bytes missing. !OS: No such file or directory !ERROR: GDKload failed: name=43, ext=theap !ERROR: BATSIGcrash: Mserver internal error (Segmentation fault), please restart. !ERROR: (One potential cause could be that your disk might be full...) Over.. ! Server not ready; skipping attempt to start client! # 22:36:07 > # 22:36:07 > Done. # 22:36:07 >

1 1

[Monetdb-developers] [bounce] [PATCH] Loop Lifted Pathfinder Sources
by Fabian Groffen 01 Nov '06

01 Nov '06

Ok, patch is a bit too big, here an URL (original message will come through later, I hope): http://dev.gentoo.org/~grobian/patches/monetdb-xquery-looplifted.patch ----- Forwarded message from monetdb-developers-bounces(a)lists.sourceforge.net ----- From: monetdb-developers-bounces(a)lists.sourceforge.net To: grobian(a)gentoo.org Subject: Your message to Monetdb-developers awaits moderator approval Date: Wed, 01 Nov 2006 13:06:01 -0800 List-Id: <monetdb-developers.lists.sourceforge.net> Your mail to 'Monetdb-developers' with the subject [PATCH] Loop Lifted Pathfinder sources Is being held until the list moderator can review it for approval. The reason it is being held: Message body is too big: 59725 bytes with a limit of 40 KB ----- End forwarded message ----- -- Fabian Groffen Gentoo on a different level

1 0

Re: [Monetdb-developers] [Monetdb-pf-checkins] pathfinder/compiler/core coreopt.brg, 1.35, 1.36 fs.brg, 1.43, 1.44
by Stefan Manegold 31 Oct '06

31 Oct '06

On Tue, Oct 31, 2006 at 09:28:29AM +0000, Jan Rittinger wrote: > Update of /cvsroot/monetdb/pathfinder/compiler/core > In directory sc8-pr-cvs7.sourceforge.net:/tmp/cvs-serv27557/core > > Modified Files: > coreopt.brg fs.brg > Log Message: > -- Aligned typing of the rule 'locpath (StepExpr, StepExpr)' (StepExpr may > e.g., be a step with an predicate) with the specification. Instead of > a typeswitch that casts all sequences of nodes into node* before applying > distinct-doc-order we now introduce a function > '#pf:distinct-doc-order-or-atomic-sequence'. It can handle the typing > without loosing the more specific type and will be replaced in the core > optimization either by a distinct-doc-order (in case of a node sequence) > or the input of the function (in case of an atomic sequence). > > (This should also solve Stefans typing problems.) yes, it does! (just like it fixed the same problem as reported in bug #1586454 "XQ: treat as") Thanks, Jan! Stefan -- | Dr. Stefan Manegold | mailto:Stefan.Manegold@cwi.nl | | CWI, P.O.Box 94079 | http://www.cwi.nl/~manegold/ | | 1090 GB Amsterdam | Tel.: +31 (20) 592-4212 | | The Netherlands | Fax : +31 (20) 592-4312 |

1 0

Re: [Monetdb-developers] [Monetdb-pf-checkins] pathfinder/tests/W3C_use_cases/XQUF/R/Tests Q7x.milS, 1.2, 1.3 Q7x.stable.out, 1.2, 1.3
by Jan Rittinger 31 Oct '06

31 Oct '06

Stefan, your rewrite may return different results (at least without schema knowledge). If you have E1/E2[P] the predicate P is evaluated on E2 only. Your rewrite changes the binding to E1/E2 and is thus equivalent to (E1/E2)[P]. The correct rewrite thus would be doc("items.xml")/items/item_tuple[itemno=1002] -> doc(\"items.xml\")/items/(for $i in item_tuple where $i/itemno=1002 return $i) which would result in the same translation as without the for loop. On 10/30/2006 10:09 PM, Stefan Manegold wrote with possible deletions: > Update of /cvsroot/monetdb/pathfinder/tests/W3C_use_cases/XQUF/R/Tests > In directory sc8-pr-cvs7.sourceforge.net:/tmp/cvs-serv14959/XQUF/R/Tests > > Modified Files: > Q7x.milS Q7x.stable.out > Log Message: > > with a path expression E and a predicate P, > Pathfinder's derives the type of > E[P] > as "node", while this could be "element" as required by some functions; > replacing > E[P] > by > for $i in E where P return $i > yields an "element". > > Added althernative versions for all tests where E[P] triggers a type error > but for $i in E where P return $i works fine. > > > Index: Q7x.stable.out > =================================================================== > RCS file: /cvsroot/monetdb/pathfinder/tests/W3C_use_cases/XQUF/R/Tests/Q7x.stable.out,v > retrieving revision 1.2 > retrieving revision 1.3 > diff -u -d -r1.2 -r1.3 > --- Q7x.stable.out 30 Oct 2006 20:54:19 -0000 1.2 > +++ Q7x.stable.out 30 Oct 2006 21:09:18 -0000 1.3 > @@ -15,7 +15,7 @@ > xquery("xml"," > do insert > <comment>This is a bargain !</comment> > -as last into exactly-one(doc(\"items.xml\")/items/item_tuple[itemno=1002]) > +as last into exactly-one(for $i in doc(\"items.xml\")/items/item_tuple where $i/itemno=1002 return $i) > ").printf(); > > xquery("xml","doc(\"items.xml\")").printf(); > > Index: Q7x.milS > =================================================================== > RCS file: /cvsroot/monetdb/pathfinder/tests/W3C_use_cases/XQUF/R/Tests/Q7x.milS,v > retrieving revision 1.2 > retrieving revision 1.3 > diff -u -d -r1.2 -r1.3 > --- Q7x.milS 30 Oct 2006 20:54:19 -0000 1.2 > +++ Q7x.milS 30 Oct 2006 21:09:17 -0000 1.3 > @@ -4,7 +4,7 @@ > xquery("xml"," > do insert > <comment>This is a bargain !</comment> > -as last into exactly-one(doc(\"items.xml\")/items/item_tuple[itemno=1002]) > +as last into exactly-one(for $i in doc(\"items.xml\")/items/item_tuple where $i/itemno=1002 return $i) > ").printf(); > > xquery("xml","doc(\"items.xml\")").printf(); > > > ------------------------------------------------------------------------- > Using Tomcat but need to do more? Need to support web services, security? > Get stuff done quickly with pre-integrated technology to make your job easier > Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 > _______________________________________________ > Monetdb-pf-checkins mailing list > Monetdb-pf-checkins(a)lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/monetdb-pf-checkins > -- Jan Rittinger Database Systems Technische Universität München (Germany) http://www-db.in.tum.de/~rittinge/

2 1

[Monetdb-developers] starting a new user...
by Romulo Goncalves 30 Oct '06

30 Oct '06

Dear developers, I created a user, schema and I set the new schema for the new user: CREATE SCHEMA "test" AUTHORIZATION "public"; CREATE USER "monet_test" WITH PASSWORD 'pass_test' NAME 'TEST USER' SCHEMA "sys"; ALTER USER "monet_test" SET SCHEMA "test"; Then I connect the new user with a mclient: [goncalve@amelia Dependencies]$ mclient -umonet_test -Ppass_test MAPI = monet_test@localhost:50000 ACTION= mapi_start_talking ERROR = !InvalidCredentialsException:checkCredentials:Invalid credentials for user 'monet_test' [goncalve@amelia Dependencies]$ What is the problem with the user creation? The authorization for the schema is public this means that any user can use the schema "test". I am right or wrong? I will appreciate some help. Best Regards, Romulo Goncalves

2 2

[Monetdb-developers] 'at'-hint URI for XRPC calls
by Ying Zhang 30 Oct '06

30 Oct '06

Hello Jens, An XRPC call needs to know the value of the 'at'-hint of its module, so that it can tell the XRPC receiver where to load the module file. The implementation of the previous XRPC syntax (import rpc-module) saves the 'at'-hint in 'apply.rpc_uri'. In the implementation of the current XRPC syntax (execute at...), this information is gone. I'm wondering if the struct PFfun_t or one of its member can be extended to include the value of the 'at'-hint. Regards, Jennie

1 1