~atcurtis/federatedx/devel : changes

~atcurtis/federatedx/devel » Changes from revision 15

From Revision 15

Rev	Summary	Authors	Date
15	fixup changes in api and import some fixes from federated fixup changes in api and import some fixes from federated	Antony T Curtis	14 years ago
14	requested_autocommit has been pushed into io implementation requested_autocommit has been pushed into io implementation	Antony T Curtis	14 years ago
13	missing change missing change	Antony T Curtis	14 years ago
12	cleanup - fix bug caused by dup var masking cleanup - fix bug caused by dup var masking	Antony T Curtis	14 years ago
11	fixup fixup	Antony T Curtis	14 years ago
10	minor fixes minor fixes	Antony T Curtis	14 years ago
9	implement support for multiple transport schemes. implement support for multiple transport schemes.	Antony T Curtis	15 years ago
8	minor refactoring minor refactoring	Antony T Curtis	15 years ago
7	refactor io instantiation refactor io instantiation	Antony T Curtis	15 years ago
6	Added missing libmysql io file Added missing libmysql io file	Patrick	15 years ago
5	Antony Curtis's Transactional patch: Antony Curtis's Transactional patch: Bug#25513 "Federated transaction failure" Experimental and fully functional patch for review/comments. What this patch provides: * Uses only 1 federated connection per server per concurrent transaction. As a result, will require significantly fewer network connections. Network connections are bound to a transaction until commit/rollback * Implements much better/more complete transactions support. * Implements primitive per-server connecton pool. connection reuse is strictly LIFO so old idle connections are permitted to time out and reconnect later when required. * Supports single-statement rollback. * Support for savepoints. Notes: 1) transaction support still requires that the remote server's storage engines have a working transaction implementation complete with savepoint support. 2) XA 2-phase commit is not yet implemented but will be trivial to add. 3) Trivial to extend in order to support other transports than mysql client protocol. 4) bulk-inserts can be implemented later as a performance improvement. 5) no limits on connection pool size - a tunable server variable may be added. 6) at shutdown, not all memory is freed. This is trivial to resolve. 7) Some indents have tab characters. Things to address later: 1) We need to use the connection pool when CREATE TABLE is performed for instances where the new federated table is using a server for which a connection already exists. 2) We need to resolve situation where CREATE TABLE requires a new network connection... Do we stall the entire server when this occurs, pick a small timeout or change LOCK_open 3) Old federated code builds SQL inconsistantly - inconsistant quoting of identifiers may open possibilities for SQL injection attack. This should be investigated.	root@hanuman	16 years ago
4	BUG #21583 Federated table returns broken strings BUG #21583 Federated table returns broken strings Added patch (modified) from Tetsuro Ikeda to set cname	Patrick Galbraith	16 years ago
3	- Fixed bug #30051 "CREATE TABLE does not connect and check ... - Fixed bug #30051 "CREATE TABLE does not connect and check existence of remote table" Modified "real_connect" to take a share and create flag to in order to not rely on any settings that are later instantiated and/or set by get_share Also, put logic in the code to not attempt this if a localhost. There's an annoying functionality that if federated tries to connect to itself during creater table, you get 1159 error (timeout) - only when local. This prevents having this functionality and is probably part of the reason it was removed. - Version 0.2 release	Patrick Galbraith	16 years ago
2	- Cleaned up license files - Cleaned up license files - Removed generated autoconf/automake files	Patrick Galbraith	16 years ago
1	Initial HG import of FederatedX 0.10 Initial HG import of FederatedX 0.10	root@arjuna	16 years ago