[#CORE-2584] Wrong results for CASE used together with GROUP BY

Firebird Core

Wrong results for CASE used together with GROUP BY

Created: 05/Aug/09 06:17 AM Updated: 05/Aug/09 08:12 PM

Return to search

Component/s:

Engine

Affects Version/s:

2.5 Beta 1, 2.5 Beta 2

Fix Version/s:

2.5 RC1

Time Tracking:

Not Specified

Issue Links:

Relate

This issue is related to:

~~CORE-501~~ Optimization problem with COALESCE

~~CORE-1343~~ Bug with a simple case and a subquery

~~CORE-2041~~ update or insert with gen_id() with wrong generator value

~~CORE-2051~~ don't work subquery in COALESCE

Planning Status:

Unspecified

Description

« Hide

Test case:

select
  rdb$field_position as a,
  case rdb$field_position when 0 then '0'
                          when 1 then '1'
                          when 2 then '2'
                          when 3 then '3'
                          when 4 then '4' end as b,
  count(*) as cnt
from rdb$relation_fields
where rdb$field_position < 5
group by 1, 2

All

Comments

Work Log

Change History

Version Control

Sort Order:

Dmitry Yemanov made changes - 05/Aug/09 06:19 AM

Field	Original Value	New Value
Assignee		Adriano dos Santos Fernandes [ asfernandes ]

Dmitry Yemanov made changes - 05/Aug/09 06:22 AM

Affects Version/s		2.5 Beta 2 [ 10300 ]
Affects Version/s	2.5 Beta 1 [ 10251 ]

[ Permalink | « Hide ]

Claudio Valderrama C. added a comment - 05/Aug/09 06:45 AM

What a funny output:

      A B CNT
======= ====== ============
      0 0 43
      1 2 42
      2 3 41
      3 4 38
      4 4 29

[ Permalink | « Hide ]

Dmitry Yemanov added a comment - 05/Aug/09 06:58 AM

This is a regression introduced after Alpha 1, probably related to other bugfixes in the grouping area happened that time.

Dmitry Yemanov made changes - 05/Aug/09 06:58 AM

Affects Version/s

2.5 Beta 1 [ 10251 ]

[ Permalink | « Hide ]

Adriano dos Santos Fernandes added a comment - 05/Aug/09 07:05 AM

Not difficult to imagine, as when this problem is happening I need to manually start finding the old related bugs, and manually run/interpret results of before/after changes. A better exercice than extend fbtcs or even run (never did that) the qm*thing.

[ Permalink | « Hide ]

Adriano dos Santos Fernandes added a comment - 05/Aug/09 09:34 AM

The bug is caused by the line in pass1.cpp with this comment:
// we do not want to reprocess last parameter later

Reason for this change is in ~~CORE-501~~ attached test and will be investigated.

Adriano dos Santos Fernandes made changes - 05/Aug/09 09:35 AM

Link

This issue is related to ~~CORE-501~~ [ ~~CORE-501~~ ]

Adriano dos Santos Fernandes made changes - 05/Aug/09 09:35 AM

Link

This issue is related to ~~CORE-1343~~ [ ~~CORE-1343~~ ]

Adriano dos Santos Fernandes made changes - 05/Aug/09 09:35 AM

Link

This issue is related to ~~CORE-2041~~ [ ~~CORE-2041~~ ]

Adriano dos Santos Fernandes made changes - 05/Aug/09 09:35 AM

Link

This issue is related to ~~CORE-2051~~ [ ~~CORE-2051~~ ]

Adriano dos Santos Fernandes made changes - 05/Aug/09 08:12 PM

Fix Version/s		2.5 RC1 [ 10362 ]
Resolution		Fixed [ 1 ]
Status	Open [ 1 ]	Resolved [ 5 ]