55 Commits

Author	SHA1	Message	Date
Jeremy Suriel	13bec8bd19	Added Encoding awareness for Ruby 1.9. Convert all data to default external encoding	2010-07-03 22:28:24 -04:00
Roger Pack	c5042772b5	add some comments to the C code. take out come compiler warnings, add some notes.	2009-04-21 02:54:29 +00:00
Roger Pack	d1433a549e	attempt to not blow up on ctrl+c during connect--that might have been a problem	2009-04-21 02:13:24 +00:00
Roger Pack	e14c8b9876	share test create db code, add a comment for the rb_thread_blocking_region helper in the C code	2009-04-18 22:55:17 +00:00
Roger Pack	f910919ffc	take out begins_with_insensitive--never liked that thing, anyway, and turned out to not be the real bug	2009-04-18 22:36:16 +00:00
Roger Pack	c235dcdf46	overcome merge conflicts	2009-04-18 22:35:02 +00:00
Roger Pack	e4bb045695	overcome a merge conflict which wasn't one	2009-04-18 22:24:31 +00:00
Roger Pack	497be9ca26	Merge branch '19_non_gvl_connect_original'	2009-04-18 22:23:17 +00:00
Roger Pack	98373d7b15	take out the rb gc for now, until I can merge in Lourens' stuff, and also add a test for multiple queries	2009-01-08 06:30:24 +00:00
Roger Pack	d2549f3907	remove test output	2008-12-20 23:07:21 +00:00
Roger Pack	9f677c3047	resolve merge conflicts	2008-12-20 23:03:56 +00:00
Hongli Lai (Phusion)	dcb1e10c16	Fix various compiler warnings.	2008-11-18 19:23:31 +01:00
Hongli Lai (Phusion)	8a758f77e3	When in Ruby 1.9, release the global interpreter lock while performing a query, so that other threads can still run.	2008-11-18 19:18:51 +01:00
Hongli Lai (Phusion)	1af96063be	Fix compilation on Ruby 1.9.	2008-11-18 18:56:40 +01:00
Lourens Naude	79f742908d	Use rb_thread_alone exclusively to determine if the query should be Thread scheduled	2008-11-01 17:12:03 +00:00
Roger Pack	8099da577d	allow the 19_non_gvl branch to work with 1.8, also fix some warnings	2008-10-31 12:37:00 -06:00
Lourens Naude	57712a64e6	Conditional schedule cleanup	2008-10-31 15:48:28 +00:00
Lourens Naude	4640893f27	Do not schedule in a single threaded environment	2008-10-31 15:09:33 +00:00
Lourens Naude	48a61dd627	Introduce Mysql#idle? && Mysql#busy? as Threaded connection pool helpers for Mysql#c_async_query	2008-10-20 01:46:03 +01:00
Lourens Naude	1af85383eb	Applied Roger's async validation patch, with a minimal test case	2008-10-19 23:17:42 +01:00
Lourens Naude	9d66a3b71e	Piggy back schedule loop on MYSQL_STATUS_READY	2008-10-19 23:04:33 +01:00
Lourens Naude	35d2545c17	Selective enable || disable GC for result retrieval with Mysql#disable_gc = true|false	2008-10-08 22:11:14 +01:00
Lourens Naude	8e6f300f9d	Introduce Mysql#reconnected?	2008-10-08 17:09:47 +01:00
Lourens Naude	f57c2a6576	Let #c_async_query support a block syntax that yields the result; ensure async in progress is negated by #get_result	2008-10-08 05:17:01 +01:00
Lourens Naude	4e5967bd24	Do not schedule && retrieve the result if #query_with_result is false	2008-10-06 23:32:43 +01:00
Roger Pack	1404c2bc44	code cleanup, since rb_thread_blocking_region actually returns the value of the function passed to it	2008-10-03 14:11:11 -06:00
Lourens Naude	85141abf09	Validate async queries	2008-10-02 01:56:10 +01:00
Roger Pack	a1d10c140b	take out some compiler warnings	2008-10-01 16:30:45 -06:00
Roger Pack	4ff863b4f9	code cleanup to avoid warnings	2008-10-01 08:09:55 -06:00
Roger Pack	ce962ebc7d	add in a function to make real_connect non blocking	2008-09-30 22:35:59 -06:00
Roger Pack	ee9cf7c47e	overcome another merge conflict	2008-09-30 17:04:06 -06:00
Roger Pack	f73ba931ef	overcome conflict	2008-09-24 11:01:14 -06:00
Lourens Naude	25a76a4d8b	Better performance without the tight loop	2008-09-10 15:02:42 +01:00
Lourens Naude	3c5dd38037	Extract thread scheduler	2008-09-10 01:26:01 +01:00
Lourens Naude	6884d454bc	Use TCP_NODELAY if supported; Remove file descriptor ops from the tight loop; Cleanup timeout assignment; Split the threaded test suite in native and c to avoid a large number of active Threads affecting results.Add experimental Mysql#socket_type which typically returns 'socket (x)', but may be helpful for debugging purposes.	2008-09-10 00:33:18 +01:00
Roger Pack	02e265072c	update the RB_UBF_DL for more recent versions of Ruby 1.9, which doesn't have it	2008-09-08 12:39:00 -06:00
Roger Pack	16d9c043c2	Merge branch 'master' into 19_full_thread_no_gil	2008-09-08 12:17:20 -06:00
Roger Pack	6c2b9ecb84	Merge branch 'master' of git@github.com:oldmoe/mysqlplus	2008-09-08 12:10:17 -06:00
Roger Pack	b1bf771d6f	add back in all_hashes method--this time hopefully with the right line lengths	2008-09-08 12:10:12 -06:00
Lourens Naude	ea2965270c	Decouple native Ruby Mysql#async_query from the C extension as per Roger's suggestion.Remove no-op c_mysql_query flag for evented tests.	2008-09-08 17:27:11 +01:00
Lourens Naude	f6f460a542	Implement Mysql#async_query as a tight loop ( 1.8 && 1.9 compatible, evented && threaded tests passing )	2008-09-08 03:56:59 +01:00
Lourens Naude	55411cd9d4	Introduce Mysql#blocking?; Set the file drescriptor for all new connections to non blocking; Add tests to the existing evented and threaded runs. ( verified 1.8 && 1.9 )	2008-09-08 01:48:47 +01:00
Lourens Naude	70c58ce67a	Allow tests to run with either the native Ruby or C based Mysql#async_query	2008-09-06 17:17:06 +01:00
oldmoe	c695d4d842	Merge branch 'master' of git@github.com:oldmoe/mysqlplus	2008-09-06 03:59:41 +02:00
oldmoe	8e1cd8ea20	fix for corrupted utf-8 results	2008-09-06 03:59:20 +02:00
Lourens Naude	951fa1d272	Access the file descriptor directly from the MySQL net structure	2008-09-06 00:01:09 +01:00
Lourens Naude	19b4b10d27	Sneak in optional timeout argument as a stub to the C version of Mysql.send_query	2008-09-05 23:55:00 +01:00
oldmoe	02e8481161	all_hashes now returns an empty array on empty results	2008-09-05 05:59:33 +02:00
Roger Pack	614568269a	rename fetch_hashes_array to process_all_hashes	2008-09-02 11:10:49 -06:00
Roger Pack	8cef960b41	also use that code for each_hash--results in 0.641s instead of 0.68s for .each_hash	2008-08-29 18:20:57 -06:00