Development talk:DB layer 2.0: Difference between revisions
From MoodleDocs
m (→PDO/ADODB) |
|||
| Line 22: | Line 22: | ||
* Next meeting. TODO: document MMM20080501, AdoDb/PDO decision, Interface + PHP documentation + Docs documentation, define tests to perform. | * Next meeting. TODO: document MMM20080501, AdoDb/PDO decision, Interface + PHP documentation + Docs documentation, define tests to perform. ADDRESS IT for... 20080508 ? | ||
[[User:Eloy Lafuente (stronk7)|Eloy Lafuente (stronk7)]] 19:00, 29 April 2008 (CDT) | [[User:Eloy Lafuente (stronk7)|Eloy Lafuente (stronk7)]] 19:00, 29 April 2008 (CDT) | ||
Revision as of 00:26, 30 April 2008
Various Notes (before 20080501)
- AdoDB, AdoDB over PDO, PDO. (my initial +1 to continue using underlying ADOdb).
- Placeholder types: :named or ? (will require PHP parsing under some DBs).
- Multiple connections supported (auto-contained DB object).
- PERF debugging available.
- Logging available (to detect wrong uses).
- Total breakage (all contrib uses)
- Total breakage (current dmllib 1.0 functions definition).
- Development mode (branch with radical replacement, some compatibility layer to minimize breakage - current dmllib instantiating new dmllib + stripping slashes ?)
- Mahara experience. ( it's also worthwhile to note that a long long time ago I also did this for elgg which involved doing the addslashes/stripslashes audit through the code base - we didn't have to do this with mahara as we had placeholders and no slashes right from the start - Penny )
- Interface example.
- dmllib 2.0 tests since the beginning.
- dmllib 2.0 documentation since the beginning.
- Classes implementation order (PG, MySQL, MSSQL, Oracle).
- More ideas.
- Next meeting. TODO: document MMM20080501, AdoDb/PDO decision, Interface + PHP documentation + Docs documentation, define tests to perform. ADDRESS IT for... 20080508 ?
Eloy Lafuente (stronk7) 19:00, 29 April 2008 (CDT)
PDO/ADODB
I think it's interesting to discuss the potential of moving to PDO instead of ADODB but I'm not convinced it needs to be in the same conversation - we should be able to switch out ADODB and use PDO instead with completely no change outside dml - surely nowhere in the code is using $db directly - should always be using execute_sql
- Yup 100% agree, underlying stuff will be completely hidden, so it's a non-priority decision. We'll be able "replace" it more or less easily or, alternatively, create new "Moodle drivers" using different internal stuff. Just prospecting your opinion about PDO. I've done some (basic) research and it seems not to be 100% ready for production under some DBs. But wanted to know if you have some experience with it (or at least you initial feeling).
