Difference between revisions of "Portable SQL/Common/DBMS Differences"
(Identifier case) |
(→Identifier case) |
||
| Line 37: | Line 37: | ||
CREATE TABLE mytable ( id INTEGER NULL ) | CREATE TABLE mytable ( id INTEGER NULL ) | ||
| − | == | + | == Identifiers == |
| − | + | Parts of this table are originally from "Building Truly Portable Database Applications in PHP": [http://www.analysisandsolutions.com/presentations/portability/slides/case.htm Column Name Case] | |
{| class="wikitable" style="text-align:center" | {| class="wikitable" style="text-align:center" | ||
| − | ! Database !! Case !! Notes | + | ! Database !! Length !! Allowed !! Case !! Notes |
|- | |- | ||
! Firebird | ! Firebird | ||
| − | | upper || preserved | + | | 31 || L,D || upper || case preserved with double quotes (") |
|- | |- | ||
! MySQL | ! MySQL | ||
| − | | preserved || | + | | 63 || L,NL,D,_,$ || preserved || |
|- | |- | ||
! Oracle | ! Oracle | ||
| − | | upper || | + | | 30 || L,NL,D,_,$,# || upper ||[http://www.dbazine.com/db2/db2-disarticles/gulutzan5 Info], max. DB name length is 8 |
|- | |- | ||
! PostgreSQL | ! PostgreSQL | ||
| − | | lower || | + | | 63 || L,NL,D,_,$ || lower || case preserved with double quotes (") |
|- | |- | ||
! SQL Server | ! SQL Server | ||
| − | | preserved || | + | | 127 || L,NL,D,_,$,@,# || preserved || [http://searchsqlserver.techtarget.com/expert/KnowledgebaseAnswer/0,289625,sid87_gci1188931,00.html Info] |
|- | |- | ||
! SQLite | ! SQLite | ||
| − | | preserved || | + | | 255 || L,NL,D,_ || preserved || |
|- | |- | ||
! Sybase ASE | ! Sybase ASE | ||
| − | | preserved || | + | | 131 || L,NL,D,_,$,@,# || preserved || max. DB and Cursor name length is 28 |
|} | |} | ||
| + | |||
| + | In general, all identifiers must start with a letter or non-latin letter; digits and other symbols as first characters are often not allowed or change their meaning. | ||
Since OpenDBX 1.1.4 the use of double quotes (") for identifiers (table and column names) is suggested. This complies to ANSI standards and is the only way to use reserved words and prevent clashes in the future if language constructs are extended by the database vendors. | Since OpenDBX 1.1.4 the use of double quotes (") for identifiers (table and column names) is suggested. This complies to ANSI standards and is the only way to use reserved words and prevent clashes in the future if language constructs are extended by the database vendors. | ||
Revision as of 15:56, 3 September 2006
Differences between implementations
NULL in column definition
This table is originally from "Building Truly Portable Database Applications in PHP": NULL
| Database | Default value | Definition allowed | Notes |
|---|---|---|---|
| Firebird | NULL | no | |
| MySQL | NULL | yes | |
| Oracle | NULL | yes | Empty strings considered NULL |
| PostgreSQL | NULL | yes | |
| SQL Server | NOT NULL | yes | |
| SQLite | NULL | yes | |
| Sybase ASE | NOT NULL | yes |
The "Definition allowed" column indicates if specifying NULL in table column definitions is allowed for CREATE TABLE statements - NOT NULL is always allowed. If "yes", the following statement doesn't generate an error:
CREATE TABLE mytable ( id INTEGER NULL )
Identifiers
Parts of this table are originally from "Building Truly Portable Database Applications in PHP": Column Name Case
| Database | Length | Allowed | Case | Notes |
|---|---|---|---|---|
| Firebird | 31 | L,D | upper | case preserved with double quotes (") |
| MySQL | 63 | L,NL,D,_,$ | preserved | |
| Oracle | 30 | L,NL,D,_,$,# | upper | Info, max. DB name length is 8 |
| PostgreSQL | 63 | L,NL,D,_,$ | lower | case preserved with double quotes (") |
| SQL Server | 127 | L,NL,D,_,$,@,# | preserved | Info |
| SQLite | 255 | L,NL,D,_ | preserved | |
| Sybase ASE | 131 | L,NL,D,_,$,@,# | preserved | max. DB and Cursor name length is 28 |
In general, all identifiers must start with a letter or non-latin letter; digits and other symbols as first characters are often not allowed or change their meaning.
Since OpenDBX 1.1.4 the use of double quotes (") for identifiers (table and column names) is suggested. This complies to ANSI standards and is the only way to use reserved words and prevent clashes in the future if language constructs are extended by the database vendors.
