Niels,

is it by accedent or by design that with mitosis & mergetable, mat.slice() is not longer used with LIMIT ?

If by design, we'd need to change (or even drop) test mat.slice_limit1.Bug-2645, that fails single-threaded since you below checkin; cf., http://monetdb.cwi.nl/testing/projects/monetdb/Candidate/sql/.mTests103/GNU....

If by accident, we'd need to investigate whether this is indeed OK ...

Stefan

On Wed, Sep 15, 2010 at 03:18:54PM +0200, Niels Nes wrote:

Changeset: d3918481778e for MonetDB URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=d3918481778e Modified Files: sql/src/test/BugTracker-2010/Tests/LIMIT_OFFSET_big-endian.Bug-2622.stable.out sql/src/test/BugTracker-2010/Tests/copy-into-too-long-string.Bug-2358.sql sql/src/test/BugTracker-2010/Tests/mat.slice_limit1.Bug-2645.stable.out sql/src/test/BugTracker-2010/Tests/with_row_number_crash.Bug-2631.stable.out Branch: Oct2010 Log Message:

approved output

[...]

diff -r b3f860694c8a -r d3918481778e sql/src/test/BugTracker-2010/Tests/mat.slice_limit1.Bug-2645.stable.out --- a/sql/src/test/BugTracker-2010/Tests/mat.slice_limit1.Bug-2645.stable.out Wed Sep 15 15:12:37 2010 +0200 +++ b/sql/src/test/BugTracker-2010/Tests/mat.slice_limit1.Bug-2645.stable.out Wed Sep 15 15:18:49 2010 +0200 @@ -112,7 +112,7 @@ % L6 # name % wrd # type % 1 # length -[ 1 ] +[ 0 ] #TRACE select cast(x as string)||'-bla-'||cast(y as string) from slice_test limit 1; % sys. # table_name % concat_concat_x # name @@ -127,7 +127,7 @@ % L6 # name % wrd # type % 1 # length -[ 1 ] +[ 0 ] #drop function GetTrace; #drop table slice_test;

[...]