lp:~sgdg/stado/stado
- Get this branch:
- bzr branch lp:~sgdg/stado/stado
Branch merges
Related bugs
Related blueprints
Branch information
Recent revisions
- 109. By Andrei Martsinchyk
-
Query planner was removing duplicate constants from the projection list.
That caused union to fail because columns was not matching.The bug was reported and fix provided by Alvin Peng
- 108. By Andrei Martsinchyk
-
SysColumn.
getColumnLegnth () was returning invalid column length for the data type VARCHAR.
The fix is provided by Alvin Peng. - 106. By Andrei Martsinchyk
-
Built-in functions are now implemented as internal array of records like "name, parameter types, result type"
If parser recognizes expression as a function, the array is searched for matching record, and expression type is taken from that record. If matching record is not found that does not necessary causing error. Main check is performed by the backend server, and there will be a problem only if function result is inserted into an intermediate table. Adding extra definitions through the config file still works, so if user is adding custom functions or using backend other then Postgres, full support for new functions can be added without code rebuild.
This commit finishes solving the problem with compiling of generated parser code because of too many defined tokens. - 105. By Andrei Martsinchyk
-
Fix the build problem, JTB generated token manager was exceeding Java 64KB limit.
So grammar was optimized, number of tokens was significantly reduced by removing function definitions matching generic rule <Identifier>(<argument> [,<argument> ]*)
This is partial fix, because Stado can not determine result type of a generic function.
The implementation of data type determining mechanism is coming. - 104. By Andrei Martsinchyk
-
Do not allow duplicate column names in a view.
If view joins multiple tables with same named columns and these column are projected, the names should be made unique by specifying column list or by specifying column aliases.
If view is queried the duplicated columns may be messed up, and ambiguity is possible. - 103. By Andrei Martsinchyk
-
New JTB 1.4.4 does not handle non-void JAVACODE productions properly.
Such production was used to define condition in CHECK definition, we did not need to parse it and sent to datanodes "as is".
Hence instead of JAVACODE use SQLComplexExpression production, so parse and reconstruct CHECK conditions. - 102. By Andrei Martsinchyk
-
Fixed bug in setting column aliases, causing possible empty column names in intermediate temp tables and step queries.
- 101. By Andrei Martsinchyk
-
Close batchStatement when resetting datanode connection.
There was a problem if batchStatement and NodeThread used different connections. The batch could not see the table created by different session and was failing. - 100. By Andrei Martsinchyk
-
Autocommit on connections to datanodes should normally be set to false.
However sometimes it may be temporary switched to true to execute specific command. However if the command failed the autocommit was not restored.
To fix the issue try ... finally block was used.
Branch metadata
- Branch format:
- Branch format 7
- Repository format:
- Bazaar repository format 2a (needs bzr 1.16 or later)
- Stacked on:
- lp:stado