Postgres Dev. - Revision history

Purcebr: /* Who We Are */

2009-05-06T04:26:27Z

Who We Are

← Older revision		Revision as of 04:26, 6 May 2009
Line 356:		Line 356:
	== Who We Are ==		== Who We Are ==

	Bryan Purcell		Bryan Purcell, purcebr[at]earlham[d0t]edu

	Mikio Takizawa		Mikio Takizawa, takizmi[at]earlham[d0t]edu

	CS430, May 2009		CS430, May 2009

Purcebr: /* Who We Are */

2009-05-06T04:25:04Z

Who We Are

← Older revision		Revision as of 04:25, 6 May 2009
Line 357:		Line 357:

	Bryan Purcell		Bryan Purcell

	Mikio Takizawa		Mikio Takizawa

	CS430, May 2009		CS430, May 2009

Purcebr at 04:24, 6 May 2009

2009-05-06T04:24:50Z

← Older revision		Revision as of 04:24, 6 May 2009
Line 353:		Line 353:

	*I skipped table alias. (miki)		*I skipped table alias. (miki)

			== Who We Are ==

			Bryan Purcell
			Mikio Takizawa
			CS430, May 2009

Purcebr at 04:23, 6 May 2009

2009-05-06T04:23:31Z

← Older revision		Revision as of 04:23, 6 May 2009
Line 1:		Line 1:
	~~Bryan Purcell~~		== Postgres Internals ==
	~~Mikio Takizawa~~

	~~== Postgres Internals ==~~		'''Bryan Purcell'''
			'''Mikio Takizawa'''
			'''CS430, May 2009'''

	We have fix a bug listed on the Postgres TODO list.		We have fix a bug listed on the Postgres TODO list.

Purcebr at 04:22, 6 May 2009

2009-05-06T04:22:18Z

← Older revision		Revision as of 04:22, 6 May 2009
Line 1:		Line 1:
			Bryan Purcell
			Mikio Takizawa

	== Postgres Internals ==		== Postgres Internals ==

Purcebr: /* Postgres Internals */

2009-05-06T04:20:11Z

Postgres Internals

← Older revision		Revision as of 04:20, 6 May 2009
Line 3:		Line 3:
	We have fix a bug listed on the Postgres TODO list.		We have fix a bug listed on the Postgres TODO list.

	''Report the schema along table name in a referential failure error message''		<pre>Report the schema along table name in a referential failure error message</pre>

	[http://wiki.postgresql.org/wiki/Todo#Referential_Integrity Described Here]		[http://wiki.postgresql.org/wiki/Todo#Referential_Integrity Described Here]

Purcebr: /* Postgres Internals */

2009-05-06T04:19:48Z

Postgres Internals

← Older revision		Revision as of 04:19, 6 May 2009
Line 1:		Line 1:
	== Postgres Internals ==		== Postgres Internals ==

	We have fix a bug listed on the Postgres TODO list.[http://wiki.postgresql.org/wiki/Todo#Referential_Integrity Described Here]		We have fix a bug listed on the Postgres TODO list.

			''Report the schema along table name in a referential failure error message''

			[http://wiki.postgresql.org/wiki/Todo#Referential_Integrity Described Here]

	The bug is catagorized in referential integrity, however we have completed a system-wide change so schema names, or namespace names, are reported along with table names in various flavors of errors. We did not have time, however, to patch the regression suite to accept the new code. When the user types ''make check'' errors are generated because the newly qualified table printout is not expected.		The bug is catagorized in referential integrity, however we have completed a system-wide change so schema names, or namespace names, are reported along with table names in various flavors of errors. We did not have time, however, to patch the regression suite to accept the new code. When the user types ''make check'' errors are generated because the newly qualified table printout is not expected.

Purcebr: /* Test Environment */

2009-05-06T04:18:37Z

Test Environment

← Older revision		Revision as of 04:18, 6 May 2009
Line 36:		Line 36:

	== Test Environment ==		== Test Environment ==

			=== Foreign Keys ===

	We have created a foreign key violation test enviroment. This this enviroment consists of two relations, where the a value in orders depends on a values in the table products. The query that causes our fk constraint failure is:		We have created a foreign key violation test enviroment. This this enviroment consists of two relations, where the a value in orders depends on a values in the table products. The query that causes our fk constraint failure is:
Line 51:		Line 53:
	DETAIL: Key (product_no)=(23) is not present in table "products".</pre>		DETAIL: Key (product_no)=(23) is not present in table "products".</pre>

	~~All of~~ the ~~foreign key constraint error cases can be found in ri_triggers.c file~~. ~~There are many other types~~ of ~~errors that will need~~ to ~~be modified, however~~, and we ~~are working on a way to automate this process~~		=== Everything Else ===

			While it's pretty easy to check the above case, it's much harder to sweep the entire system recreating errors to test our reformatting.

			Instead of creating these cases yourself, run

			<pre>make check</pre>

			at the root directory to get a diff of the old table names, and the newly formatted table names. The make check will end in error because we have not patched the regression suite.

	== Automation ==		== Automation ==

Purcebr: /* Automation */

2009-05-06T04:15:32Z

Automation

← Older revision		Revision as of 04:15, 6 May 2009
Line 55:		Line 55:
	== Automation ==		== Automation ==

	The bug description on the mailing list seemed to find this fix more tedious than outright difficult. Now that we've seen the fix work, it seems like a good idea to try to ~~automate corrections across~~ the ~~whole system~~.		The bug description on the mailing list seemed to find this fix more tedious than outright difficult. Now that we've seen the fix work, it seems like a good idea to try to figure out all the relevant places to fix.

	<pre>		<pre>

Purcebr: /* Postgres Internals Notes */

2009-05-06T04:14:57Z

Postgres Internals Notes

← Older revision		Revision as of 04:14, 6 May 2009
Line 1:		Line 1:
	== Postgres Internals ~~Notes~~==		== Postgres Internals ==

	=== ~~Current Efforts~~ ===		We have fix a bug listed on the Postgres TODO list.[http://wiki.postgresql.org/wiki/Todo#Referential_Integrity Described Here]
			The bug is catagorized in referential integrity, however we have completed a system-wide change so schema names, or namespace names, are reported along with table names in various flavors of errors. We did not have time, however, to patch the regression suite to accept the new code. When the user types ''make check'' errors are generated because the newly qualified table printout is not expected.

			The following includes some notes from the project along with the procedure for testing.

			=== Notes ===

	<pre>get_namespace_name(oid)</pre>		<pre>get_namespace_name(oid)</pre>
Line 20:		Line 25:
	ri_triggers.c houses all cases where fk constraints fail. There are 5 or 6 variations. The primary error message is the same in each case, and a detail section gets more specific about why the constraint blocked the SQL action.		ri_triggers.c houses all cases where fk constraints fail. There are 5 or 6 variations. The primary error message is the same in each case, and a detail section gets more specific about why the constraint blocked the SQL action.

	~~Fixed~~ error report in ri_triggers.c :		Sample error report generated from an fixed version of ri_triggers.c :

	<pre>		<pre>
Line 29:		Line 34:
	3491 strcat( strcat(get_namespace_name(fk_rel->rd_rel->relnamespace), "."), RelationGetRelationName(fk_rel)))));		3491 strcat( strcat(get_namespace_name(fk_rel->rd_rel->relnamespace), "."), RelationGetRelationName(fk_rel)))));
	</pre>		</pre>

	The problem we have to figure out here is how to combine the namespace name and table name into a single variable. Adding another %s will require changing each localization file, along with the source code. Making such a substantial change probably won't be possible.

	== Test Environment ==		== Test Environment ==