Re: [Monetdb-developers] [Monetdb-pf-checkins] pathfinder/compiler/mil milgen.brg, Feb2009, 1.201.2.10, 1.201.2.11
Is this a reason to rebuild the release? Yesterday I built release candidates and I don't really want to do that again if I can avoid it. Jan Rittinger wrote:
Update of /cvsroot/monetdb/pathfinder/compiler/mil In directory 23jxhf1.ch3.sourceforge.com:/tmp/cvs-serv7388/mil
Modified Files: Tag: Feb2009 milgen.brg Log Message: -- Fix alignment problem in the MIL generation of the slice() operator.
Scary, but true: Some rules in the MIL generation require the head values to start with 0@0 as they add new columns with offset 0.
U milgen.brg Index: milgen.brg =================================================================== RCS file: /cvsroot/monetdb/pathfinder/compiler/mil/milgen.brg,v retrieving revision 1.201.2.10 retrieving revision 1.201.2.11 diff -u -d -r1.201.2.10 -r1.201.2.11 --- milgen.brg 18 Mar 2009 09:16:50 -0000 1.201.2.10 +++ milgen.brg 24 Apr 2009 10:12:15 -0000 1.201.2.11 @@ -4708,7 +4708,11 @@ assgn (var (tmp->name), slice (var (env_at (L(p)->env, i).mvar->name), lit_int (p->sem.slice.low), - lit_int (p->sem.slice.high)))); + lit_int (p->sem.slice.high))), + /* scary but true -- we need to let the seqbases start + with 0@0 as otherwise alignment problems arise */ + assgn (var (tmp->name), + tmark (var (tmp->name), lit_oid (0))));
env_add (p->env, env_at (L(p)->env, i).col,
------------------------------------------------------------------------------ Crystal Reports - New Free Runtime and 30 Day Trial Check out the new simplified licensign option that enables unlimited royalty-free distribution of the report engine for externally facing server and web deployment. http://p.sf.net/sfu/businessobjects _______________________________________________ Monetdb-pf-checkins mailing list Monetdb-pf-checkins@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/monetdb-pf-checkins
-- Sjoerd Mullender
It affects queries using subsequence() with two more conditions: 1.) some items at the start are skipped and 2.) the result of subsequence is used as input for other expressions. I think the problem can also be avoided by using conditions on positions instead (for $a at $pos in ... where $pos > 10). So it's your decision... On Apr 24, 2009, at 12:51, Sjoerd Mullender wrote: > Is this a reason to rebuild the release? > > Yesterday I built release candidates and I don't really want to do > that > again if I can avoid it. > > Jan Rittinger wrote: >> Update of /cvsroot/monetdb/pathfinder/compiler/mil >> In directory 23jxhf1.ch3.sourceforge.com:/tmp/cvs-serv7388/mil >> >> Modified Files: >> Tag: Feb2009 >> milgen.brg >> Log Message: >> -- Fix alignment problem in the MIL generation of the slice() >> operator. >> >> Scary, but true: Some rules in the MIL generation require the head >> values to start with 0@0 as they add new columns with offset 0. >> >> >> U milgen.brg >> Index: milgen.brg >> =================================================================== >> RCS file: /cvsroot/monetdb/pathfinder/compiler/mil/milgen.brg,v >> retrieving revision 1.201.2.10 >> retrieving revision 1.201.2.11 >> diff -u -d -r1.201.2.10 -r1.201.2.11 >> --- milgen.brg 18 Mar 2009 09:16:50 -0000 1.201.2.10 >> +++ milgen.brg 24 Apr 2009 10:12:15 -0000 1.201.2.11 >> @@ -4708,7 +4708,11 @@ >> assgn (var (tmp->name), >> slice (var (env_at (L(p)->env, i).mvar- >> >name), >> lit_int (p->sem.slice.low), >> - lit_int (p->sem.slice.high)))); >> + lit_int (p->sem.slice.high))), >> + /* scary but true -- we need to let the >> seqbases start >> + with 0@0 as otherwise alignment problems >> arise */ >> + assgn (var (tmp->name), >> + tmark (var (tmp->name), lit_oid (0)))); >> >> env_add (p->env, >> env_at (L(p)->env, i).col, >> >> >> ------------------------------------------------------------------------------ >> Crystal Reports - New Free Runtime and 30 Day Trial >> Check out the new simplified licensign option that enables unlimited >> royalty-free distribution of the report engine for externally >> facing >> server and web deployment. >> http://p.sf.net/sfu/businessobjects >> _______________________________________________ >> Monetdb-pf-checkins mailing list >> Monetdb-pf-checkins@lists.sourceforge.net >> https://lists.sourceforge.net/lists/listinfo/monetdb-pf-checkins > > > -- > Sjoerd Mullender > > ------------------------------------------------------------------------------ > Crystal Reports - New Free Runtime and 30 Day Trial > Check out the new simplified licensign option that enables unlimited > royalty-free distribution of the report engine for externally > facing > server and web deployment. > http://p.sf.net/sfu/businessobjects_______________________________________________ > Monetdb-pf-checkins mailing list > Monetdb-pf-checkins@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/monetdb-pf-checkins -- Jan Rittinger Lehrstuhl Datenbanken und Informationssysteme Wilhelm-Schickard-Institut für Informatik Eberhard-Karls-Universität Tübingen http://www-db.informatik.uni-tuebingen.de/team/rittinger
participants (2)
-
Jan Rittinger
-
Sjoerd Mullender