mike [Fri, 21 Sep 2007 11:40:50 +0000 (11:40 +0000)]
Move statement about NLA.
mike [Fri, 21 Sep 2007 11:31:09 +0000 (11:31 +0000)]
Change LC URL *sigh*
mike [Mon, 6 Aug 2007 15:59:05 +0000 (15:59 +0000)]
New
mike [Mon, 6 Aug 2007 15:55:03 +0000 (15:55 +0000)]
Release 1.2.1
mike [Mon, 6 Aug 2007 15:54:48 +0000 (15:54 +0000)]
Constructors are public.
mike [Tue, 3 Jul 2007 17:04:18 +0000 (17:04 +0000)]
1.2.1
mike [Tue, 3 Jul 2007 17:04:09 +0000 (17:04 +0000)]
Credit NLA.
mike [Tue, 3 Jul 2007 17:03:15 +0000 (17:03 +0000)]
New
mike [Tue, 3 Jul 2007 16:54:50 +0000 (16:54 +0000)]
Release 1.2
mike [Tue, 3 Jul 2007 16:42:02 +0000 (16:42 +0000)]
Version 1.2
mike [Tue, 3 Jul 2007 16:41:04 +0000 (16:41 +0000)]
New
mike [Tue, 3 Jul 2007 16:40:41 +0000 (16:40 +0000)]
Fail a sortspec of zero keys.
mike [Tue, 3 Jul 2007 16:40:11 +0000 (16:40 +0000)]
PrefixNode propagates sort keys.
mike [Tue, 3 Jul 2007 16:39:47 +0000 (16:39 +0000)]
Append many sorting-related tests.
mike [Tue, 3 Jul 2007 16:03:00 +0000 (16:03 +0000)]
toXCQL() renders sort-keys, if any.
mike [Tue, 3 Jul 2007 15:53:52 +0000 (15:53 +0000)]
Do not omit section 10
mike [Tue, 3 Jul 2007 15:41:35 +0000 (15:41 +0000)]
Boolean nodes are generated with an appropriate ModifierSet. It's
horribly wrong that the MSs need to know the name of the boolean.
mike [Tue, 3 Jul 2007 13:41:24 +0000 (13:41 +0000)]
Implement new, wider toXCQL() API.
Render sort keys, if any.
mike [Tue, 3 Jul 2007 13:40:13 +0000 (13:40 +0000)]
Implement new, wider toXCQL() API.
mike [Tue, 3 Jul 2007 13:36:03 +0000 (13:36 +0000)]
Primary toXCQL() API now includes a Vector of sort keys.
A thin-layer presents the old API.
Add renderSortKeys() method.
mike [Tue, 3 Jul 2007 13:34:30 +0000 (13:34 +0000)]
New method parseTopLevelPrefixes() handles both initial prefixes and a
sortby clause if any.
parseQuery() breaks on SORTBY as well as ")" and EOF.
parsePrefix() now takes an additional boolean argument topLevel: if
true, then recursion is to parseTopLevelPrefixes() rather than
parseQuery().
mike [Tue, 3 Jul 2007 13:30:42 +0000 (13:30 +0000)]
Add "sortby" keyword.
mike [Tue, 3 Jul 2007 13:30:18 +0000 (13:30 +0000)]
toXCQL() refactored as layer over new underlyingToXCQL()
Add sortKeyToXCQL() as another layer over underlyingToXCQL()
mike [Tue, 3 Jul 2007 13:29:34 +0000 (13:29 +0000)]
Simplify.
mike [Tue, 3 Jul 2007 12:56:29 +0000 (12:56 +0000)]
Include CQLSortNode.class
mike [Tue, 3 Jul 2007 12:55:56 +0000 (12:55 +0000)]
New
mike [Fri, 29 Jun 2007 15:56:48 +0000 (15:56 +0000)]
Rolling
mike [Fri, 29 Jun 2007 15:42:17 +0000 (15:42 +0000)]
New
mike [Fri, 29 Jun 2007 15:42:02 +0000 (15:42 +0000)]
Add three new queries to test lexically complex relations.
mike [Fri, 29 Jun 2007 15:38:56 +0000 (15:38 +0000)]
Support for "==" relation (TT_EQEQ)
mike [Fri, 29 Jun 2007 15:26:29 +0000 (15:26 +0000)]
isResultSetIndex() recognises cql.* indexes as well as srw.*
toCQL() recognises cql.serverChoice as well as srw.serverChoice.
mike [Fri, 29 Jun 2007 15:24:39 +0000 (15:24 +0000)]
Comment.
mike [Fri, 29 Jun 2007 15:24:26 +0000 (15:24 +0000)]
New constructor that takes a "compat" argument indicating whether to
implement v1.1, v1.2 (the default) or v1.1 with sorting.
Default relation generated is "=" under v1.2.
Add new "-1" command-line argument to test-harness for v1.1.
mike [Fri, 29 Jun 2007 15:22:52 +0000 (15:22 +0000)]
Towards v1.2
mike [Fri, 29 Jun 2007 15:22:27 +0000 (15:22 +0000)]
Expect = as default relation rather than scr
mike [Fri, 29 Jun 2007 15:10:16 +0000 (15:10 +0000)]
Expect cql.serverChoice instead of srw.serverChoice
mike [Fri, 29 Jun 2007 13:19:28 +0000 (13:19 +0000)]
New
mike [Fri, 29 Jun 2007 13:17:04 +0000 (13:17 +0000)]
Remove archive directory
mike [Fri, 29 Jun 2007 13:10:58 +0000 (13:10 +0000)]
Release 1.0
mike [Fri, 29 Jun 2007 13:06:07 +0000 (13:06 +0000)]
Describe runcanon
mike [Fri, 29 Jun 2007 13:05:12 +0000 (13:05 +0000)]
New
mike [Fri, 29 Jun 2007 12:54:05 +0000 (12:54 +0000)]
Remove obsolete keywords! Whoop!
mike [Fri, 29 Jun 2007 12:53:03 +0000 (12:53 +0000)]
Remove op() method.
mike [Fri, 29 Jun 2007 12:52:45 +0000 (12:52 +0000)]
Remove op() method, use ms.toCQL()
mike [Fri, 29 Jun 2007 12:48:21 +0000 (12:48 +0000)]
Remove op()
mike [Fri, 29 Jun 2007 12:27:08 +0000 (12:27 +0000)]
Remove old proximity-specific parsing code.
mike [Fri, 29 Jun 2007 12:26:44 +0000 (12:26 +0000)]
Remove extraneous import.
mike [Fri, 29 Jun 2007 12:15:05 +0000 (12:15 +0000)]
Throw away a lot of special-case code -- hurrah! :-)
The base BooleanNode code covers pretty much everything now.
mike [Fri, 29 Jun 2007 11:56:47 +0000 (11:56 +0000)]
Treat PROX the same as AND, OR and NOT. Nice.
gatherModifiers() now uses matchSymbol() instead of duplicating code.
mike [Fri, 29 Jun 2007 11:56:06 +0000 (11:56 +0000)]
Pass-through constructor.
mike [Fri, 29 Jun 2007 11:55:45 +0000 (11:55 +0000)]
Temporarily change the token "ordered" to "xordered" so that the word
is not recognised as special. Soon, nearly all keywords (including
this one) will be removed.
mike [Fri, 29 Jun 2007 11:54:56 +0000 (11:54 +0000)]
Comments
mike [Fri, 29 Jun 2007 11:48:04 +0000 (11:48 +0000)]
Add "make test" rule.
mike [Fri, 29 Jun 2007 11:47:44 +0000 (11:47 +0000)]
v1.1 XCQL for proximity modifiers
mike [Fri, 29 Jun 2007 11:47:25 +0000 (11:47 +0000)]
Clearer diagnostics when output differs.
mike [Fri, 29 Jun 2007 10:25:38 +0000 (10:25 +0000)]
New gatherModifiers() method, used to implement boolean modifiers.
parseTerm() modified to use gatherModifiers() for relation modifiers.
mike [Fri, 29 Jun 2007 10:24:24 +0000 (10:24 +0000)]
Use new three-argument forms of boolean-node constructors.
mike [Fri, 29 Jun 2007 10:24:03 +0000 (10:24 +0000)]
Removed ModifierSet, since base class now has this.
Removed opXCQL() method, as base class does this for us.
mike [Fri, 29 Jun 2007 10:23:02 +0000 (10:23 +0000)]
Remove addModifier() methods with setModifiers().
Client code should directly build a ModifierSet, then set it.
mike [Fri, 29 Jun 2007 10:22:12 +0000 (10:22 +0000)]
Constructor accepts a ModifierSet
mike [Fri, 29 Jun 2007 10:21:30 +0000 (10:21 +0000)]
Boolean node now carries a ModifierSet.
Base constructor provided for subclasses to invoke.
mike [Fri, 29 Jun 2007 10:20:41 +0000 (10:20 +0000)]
Add single-argument addModifier.
mike [Fri, 29 Jun 2007 10:20:09 +0000 (10:20 +0000)]
Fix single-argument constructorm, which no-opped.
mike [Thu, 28 Jun 2007 17:33:56 +0000 (17:33 +0000)]
New
mike [Thu, 28 Jun 2007 00:24:48 +0000 (00:24 +0000)]
Parse complex relation modifiers.
(Getting the lexeme out of the lexer involves sticking my fingers down
its throat, which is now done in two places. Should be refactored
into the lexer itself.)
mike [Thu, 28 Jun 2007 00:23:43 +0000 (00:23 +0000)]
Three-parameter version of addModifier()
mike [Thu, 28 Jun 2007 00:23:17 +0000 (00:23 +0000)]
toXCL() emits modifiers with "comparison" rather than "relation"
mike [Thu, 28 Jun 2007 00:22:29 +0000 (00:22 +0000)]
new
mike [Thu, 28 Jun 2007 00:22:18 +0000 (00:22 +0000)]
New
mike [Thu, 28 Jun 2007 00:00:53 +0000 (00:00 +0000)]
Remove relation keywords TT_ANY, TT_ALL, and TT_EXACT.
Remove relation-modifier keywords TT_RELEVANT, TT_FUZZY, TT_STEM,
TT_SCR and TT_PHONETIC.
Remove unnecessary fascistic checks.
I feel so free!
mike [Thu, 28 Jun 2007 00:00:31 +0000 (00:00 +0000)]
Remove relation keywords TT_ANY, TT_ALL, and TT_EXACT.
Remove relation-modifier keywords TT_RELEVANT, TT_FUZZY, TT_STEM,
TT_SCR and TT_PHONETIC.
I feel so free!
mike [Wed, 27 Jun 2007 23:36:19 +0000 (23:36 +0000)]
New
mike [Wed, 27 Jun 2007 22:44:40 +0000 (22:44 +0000)]
UnknownQualifierException becomes UnknownIndexException
mike [Wed, 27 Jun 2007 22:44:13 +0000 (22:44 +0000)]
Rolling
mike [Wed, 27 Jun 2007 22:41:41 +0000 (22:41 +0000)]
Renamemv UnknownQualifierException.java to UnknownIndexException.java
mike [Wed, 27 Jun 2007 22:39:55 +0000 (22:39 +0000)]
'qualifier' replaced by 'index' throughout
mike [Wed, 27 Jun 2007 22:16:08 +0000 (22:16 +0000)]
Rolling
mike [Wed, 27 Jun 2007 22:14:46 +0000 (22:14 +0000)]
addModifier() now uses broader ModifierSet API.
mike [Wed, 27 Jun 2007 22:14:03 +0000 (22:14 +0000)]
addModifier() now accepts comparison as well as type and value.
mike [Wed, 27 Jun 2007 17:24:04 +0000 (17:24 +0000)]
Updated to new relation-modifier format
mike [Wed, 27 Jun 2007 17:18:54 +0000 (17:18 +0000)]
Single-argument constructor sets type rather than value.
mike [Wed, 27 Jun 2007 17:18:00 +0000 (17:18 +0000)]
addModifier() correctly calls changed ModifierSet API.
mike [Wed, 27 Jun 2007 17:05:05 +0000 (17:05 +0000)]
Use new Modifier objects for PQF translation rather than the old
Vector hack.
mike [Wed, 27 Jun 2007 17:02:01 +0000 (17:02 +0000)]
Simplify getModifiers.
mike [Wed, 27 Jun 2007 17:01:47 +0000 (17:01 +0000)]
Build Modifier.java
mike [Wed, 27 Jun 2007 17:01:38 +0000 (17:01 +0000)]
Use new Modifier class instead of two-element Vector hack.
mike [Wed, 27 Jun 2007 17:01:23 +0000 (17:01 +0000)]
New
mike [Tue, 26 Jun 2007 16:43:41 +0000 (16:43 +0000)]
Add new -d option to test-harness to turn on debugging.
mike [Tue, 26 Jun 2007 16:36:50 +0000 (16:36 +0000)]
Fix rendering of default-context nodes with no named prefix.
mike [Thu, 21 Jun 2007 14:54:32 +0000 (14:54 +0000)]
Failure to compile a query counts as a failed test!
mike [Thu, 21 Jun 2007 10:15:58 +0000 (10:15 +0000)]
Fixed for modified query.
mike [Thu, 21 Jun 2007 10:10:32 +0000 (10:10 +0000)]
Emit "=== end ===" after comparison.
mike [Thu, 21 Jun 2007 10:00:24 +0000 (10:00 +0000)]
Changed proximity syntax for v1.1/1.2
Added other queries with boolean modifiers
Tweak comments and context-set URIs
mike [Tue, 19 Jun 2007 08:47:25 +0000 (08:47 +0000)]
New, downloaded from LC
mike [Thu, 7 Jun 2007 16:17:28 +0000 (16:17 +0000)]
Final report of success count and percentage.
mike [Thu, 7 Jun 2007 16:17:04 +0000 (16:17 +0000)]
New "no thinking" introduction.
Note on shared code in runtests/mkanswers.
Describe showtest.
mike [Thu, 7 Jun 2007 16:16:15 +0000 (16:16 +0000)]
PATH and CLASSPATH settings so that no installation is needed.
mike [Thu, 7 Jun 2007 16:15:22 +0000 (16:15 +0000)]
Revert the previous change, because it broke the regression test.
This means that terms like "3d" will, as before, be incorrectly lexed
as the number "3" followed by the word "d", but that's how every
release until now has been, so we can obviously live with that.
The regression test was also broken by the previous change supporting
floating-point numbers, but I fixed that by having render() return
only the integer part when that's numerically equal to the full
banana. A similar lexical hack might be the way forward with the
other reverted change, but right now just making everything work the
way it ought to top priority.
mike [Thu, 7 Jun 2007 16:04:50 +0000 (16:04 +0000)]
Up to date description.
mike [Wed, 6 Jun 2007 15:13:41 +0000 (15:13 +0000)]
Comment.