32 Коммитов

Автор	SHA1	Сообщение	Дата
Mark Reid	7d06f24fdb	Update Makefile for compatibility with CDH4	2013-07-10 11:42:18 -03:00
Mark Reid	6600e15f82	Refactored to use HadoopDriver.java	2013-06-17 17:10:52 -03:00
Mark Reid	64b16db3d9	Make Makefile work on mango-gw	2013-06-17 14:10:26 -03:00
Mark Reid	45a5544e6e	First stab	2013-06-17 11:54:55 -03:00
Mark Reid	7b606548d9	Update makefile to work on the 'peach' cluster	2013-06-17 10:14:33 -03:00
Taras Glek	13e9e29df7	a possibly-working json.dumps	2013-05-01 18:03:08 -07:00
Taras Glek	62ae5c759f	run tests via our wrapper	2013-05-01 14:35:40 -07:00
Gregory Szorc	d0ff9d698e	Move count_fhr_facets.py into fhr script subdir	2013-04-23 16:25:41 -07:00
Gregory Szorc	35a0136f21	Move common modules to jydoop directory	2013-04-22 13:17:15 -07:00
Benjamin Smedberg	6a632f7bb6	Separate the type and value classes so that values can contain lists and dicts (and are not sortable) while keys cannot contain those and are sortable.	2013-04-19 09:35:57 -07:00
Taras Glek	aa0afd15c6	added a way to pull resources out of driver.jar	2013-04-18 16:28:35 -07:00
Benjamin Smedberg	2fdda52ddd	Rename pydoop->jydoop because there is already a pretty well-known project called pydoop.	2013-04-08 16:07:14 -07:00
Benjamin Smedberg	3ffed3736e	Use Jackson instead of Jyson for speed. See also TODO for future speed improvements in JacksonWrapper.loads.	2013-04-06 09:35:28 -07:00
Taras Glek	56fa7e30ce	removed from leftovers from proof-of-concept days	2013-04-04 11:33:14 -07:00
Benjamin Smedberg	086147ca31	Ship all the scripts in driver.jar and select which one to run at runtime (preparation for a standard library of pydoop python scripts).	2013-04-04 09:27:20 -07:00
Benjamin Smedberg	c2a23923a8	Implement tests for TypeWritable and its implementation of RawComparator.	2013-04-03 15:18:55 -07:00
Benjamin Smedberg	83c5860c41	Use jyson again. The builtin json module that comes with jython is dog-slow: 9 seconds to parse a single telemetry record!	2013-03-29 11:47:35 -04:00
Benjamin Smedberg	a798248a5f	Fix makefile to use the new package names and Text is no longer correct.	2013-03-25 15:38:35 -04:00
Benjamin Smedberg	4aeb87af81	Move the Java into the correct directory structure expected by a java compiler, and give it a more correct package name. Add a Writeable class which is used to efficiently serialize just the data we might actually want in a python key/value, so that we can now emit tuples and integers.	2013-03-22 03:27:37 +01:00
Benjamin Smedberg	39475c74ef	Use wget -c per taras	2013-03-21 10:38:37 +01:00
Benjamin Smedberg	c07e0a22e1	Fix misspelling.	2013-03-21 09:59:45 +01:00
Benjamin Smedberg	13d8b684c6	Actually remove the HDFS driver which we don't need.	2013-03-21 09:57:42 +01:00
Benjamin Smedberg	6a24e3c8b0	Use curl and > to avoid redownload issues.	2013-03-21 09:57:05 +01:00
Benjamin Smedberg	4ecb4b8a90	Use the proper jython-standalone.jar, which makes Jyson unnecessary. Also pull akela.	2013-03-21 09:57:05 +01:00
Benjamin Smedberg	62f16e8d75	Make the path of hbase configurable, and make it easier to build.	2013-03-21 09:50:31 +01:00
Taras Glek	cd6fc4b416	Some more docs	2013-03-18 08:45:52 -07:00
Taras Glek	fdb36db1ed	anr script	2013-03-13 17:45:35 -07:00
Taras Glek	aa524c8206	Makefile fix, use compression for hbase map output	2013-03-13 16:21:29 -07:00
Taras Glek	237b2b5e21	some cleanup	2013-03-13 12:12:53 -07:00
Taras Glek	c9f56fa339	got jython running against hbase too	2013-03-12 23:33:49 -07:00
Taras Glek	b162584b2e	first successful map/reduce with json	2013-03-06 21:12:37 -08:00
Taras Glek	cb18d8598d	got wordcount to run partially in python	2013-03-06 12:02:18 -08:00