Done

This file contains completed tasks.
See also:
  1. 1998-06-12 Bug: true branch of <if> failing.
    (1998-06-11)
    -- This turned out to be due to the fact that, unlike java.util.Enumeration, a NodeEnumerator skips the first item if you start with getNext instead of getFirst.
  2. 1998-06-12 Check: Loops with NodeEnumerator
    Loops involving a NodeEnumerator have to start with getFirst -- getNext skips the first element (unlike a java.util.Enumeration)! (1998-06-12)
  3. 1998-06-12 Bug: testHandler
    <test>foo</test> returns 11 -- extra put, perhaps. (1998-06-11)
    -- Missing return, actually, after the test for null attribute list.
  4. 1998-06-15 Bug: entity substitution
    &date; not getting substituted (i.e. not getting looked up) in running text, but substituted correctly in attributes. (1998-06-11)
    -- Caused by AbstractHandler not checking for NodeType == ENTITY.
  5. 1998-06-15 Bug: nested <if>
    The correct branch is selected, but appears twice in the output. (1998-06-12)
    -- [copying a node that actually has children should just do a putNode and not recurse.]
  6. 1998-06-15 Bug: <if text>
    any element in the condition appears to be considered true. (1998-06-12)
    -- Incorrect dispatching: wasn't dispatching if no subtype attr. This meant that text, specifically, wasn't handled.
  7. 1998-06-15 Hole: Filter parse tree output
    parse tree input & output should use serialized object streams. (1998-06-15)
    -- The -n option was added to permit parsing but not expanding an input file.
  8. 1998-06-16 Hole: get/set)IndexValue
    Should do this using utils in org.risource.dps.aux.Index.
  9. 1998-06-17 Hole: org.risource.dps.active as DOM
    Fairly soon we will want to expand ParseTree* to a full DOM implementation, and make it completely independent from the implementations in org.risource.dom.
    -- completed 6/17 with some bugs remaining, and no impl. of Document.
  10. 1998-06-17 Impr: AbstractActiveNode deepCopy
    deepCopy should use a method in the abstract parent to copy the children. The reason it doesn't now is that the existing method meant for that purpose is in org.risource.dom.AbstractNode, and doesn't constructive active nodes. It uses clone, and it's broken.
    -- fixed with the re-parenting of the ParseTree classes.
  11. 1998-06-18 Bug: Bugs with independent ParseTree DOM
    After making ParseTreeElement, etc. descend from ParseTreeNode, empty elements are not identified, and <test match> produces odd results. Start tags are getting replaced by end tags in some cases! (1998-06-17)
    -- isEmptyElement is never getting set. Do it in AbstractParser's createActiveElement and correctActiveElement.
  12. 1998-06-18 Impr: EntityHandler needed
    having a specific handler for expandable entities would help a lot. It is expected that this would result in a substantial speedup. (1998-06-15)
    -- In all about 5 seconds (15%) off parsing from other improvements, and 6 off processing from this one.
  13. 1998-06-18 Impr: TextHandler needed
    -- no noticable speed improvement.
  14. 1998-06-18 Impr: Remove debugging code from inner loops.
    -- Removing debug calls in processNode and additionalAction saves another 30%! (sick.)
  15. 1998-06-18 Hole: PUT_NODE action code needed
    Passive node with no children -- just put it. (1998-06-18)
  16. 1998-06-18 Hole: ParseTreeNode must allow extra action on addChild, insertBefore, and initialization.
    This will allow insertion methods to be extended. Only the protected dispatch method doInsert needs to be extended. (1998-06-19)
  17. 1998-06-18 Hole: Tagset and Handler must be generic.
    Sometimes we may need them to be elements, sometimes declarations. (1998-06-18)
  18. 1998-06-18 Hole: Tagset and EntityTable must be lists as well.
    The items defined must be their content, not their attributes. This will greatly simplify the loading of tagsets, DTD's, etc. (1998-06-19)
  19. 1998-06-22 Impr: Define syntax flags in org.risource.dps.Syntax
    (They are currently defined in BasicTagset, which makes them mostly inaccessible.) (1998-06-18)
  20. 1998-06-23 Bug: entities and attributes broken
    Something broke when we went to values not being children. Values are set, but cannot be retreived. Several places seem to still be using getChildren instead of getValue. (1998-06-23)
    -- tested old value for null instead of new, in putValue.
  21. 1998-06-23 Bug: <test match> produces odd results.
    <test match> produces odd results sometimes in t2.if (1998-06-17)
    -- due to attributes being broken, of course.
  22. 1998-06-24 Bug: Nested <if> loses <then> content.
    The children of a second-level then are getting lost instead of copied. (1998-06-23)
    -- declaring int action instead of int actionCode
  23. 1998-06-24 Impr: <test> should be ended by <then>
    (1998-06-18)
  24. Check: copying
    Non-processed parse trees (e.g. <else>) should never be copied. Copying should use putNode if the node actually has children; recursion if the children are unparsed. (1998-06-15)
  25. 1998-06-25 Impr: Dispatching: create fewer instances
    Dispatching should create a new handler instance only when necessary; normally it should cache instances with standard attributes. (1998-06-15)
    -- handle(element) method in place; usage needs to be checked.
  26. 1998-06-25 Query: <test zero> non-number</> = ?
    Not clear whether non-numerics should test as zero or not. The most consistant thing would be for whitespace to return true, non-whitespace to return false. (1998-06-15) -- At the moment, whitespace is zero.
  27. 1998-06-26 Bug: Empty actor def's fail.
    (1998-06-25)
    -- Empty def. handled by defining as null string.
  28. 1998-06-27 Bug: Actors defined inside <if> fail.
    One can no longer expect an actor defined inside quotes to be visible there (because of pre-parsing), but it should be visible outside. (1998-06-25)
    -- ParseTreeText wasn't getting cloned correctly; a hangover from when it descended from BasicText.
  29. 1998-06-27 Bug: Parsing: Lots of trouble with implicit end.
    <tr> ends the previous <th>, but neither the expected </tr> nor the <tr> show up. An entity before an inserted </td> also causes havoc -- it turns into a comment! (probably bad node type). Missing </then> also loses. (1998-06-26)
    -- Not putting token back into next in advanceParser.
  30. 1998-06-27 Goal: (7-1) Able to process slides
    Be able to process .../Reviews/slides-199807.html in time to show it at the review. Requires <actor> (1998-06-24)
  31. 1998-07-05 Impr: Tagset should not extend DOMFactory.
    Besides; DOMFactory is going away. (1998-06)
  32. 1998-07-07 Impr: Pre-load tagset and handler classes.
    Probably best done by having explicit tagset and handler cache classes. (1998-07-07)
  33. 1998-07-09 Hole: PIA transition from old to new interp.
    Agents need to know which interpretor (and tagset, of course) they should be using. Hooks can be distinguished by type. (1998-06-18)
  34. 1998-07-08 Hole: PIA Content in push mode.
    org.risource.pia.Content currently operates in pull mode, expecting to read from an input stream. For the DPS it needs to run in push mode, giving the DPS a suitable output stream and letting it run. (1998-06-18)
    -- (Actually, Content always did the right thing; it was just StreamContent that didn't. The alternative, though, never handled insertions correctly.)
  35. 1998-07-15 Impr: <test iftrue/iffalse> not needed.
    The iftrue and iffalse attributes are both unnecessary and unused. (1998-07-15)
  36. 1998-07-17 Impr: Move Processor implementations to subdir.
    (1998-07-08)
    -- moved to org.risource.dps.process
  37. 1998-07-20 Bug?: XML PI's need to end in `?>'
    present parser ends them with `>'. (1998-07-16)
    -- This may be a misunderstanding. Certainly in generic SGML, PI's end with plain `>', and even some XML documentation so specifies. The XML identifier, however, definitely does require a ?.

  38. Wish: ParseTreeGenericNode
    Eventually there should be a ParseTreeGenericNode, basically Token, so that we can build trees with arbitrary node types.
  39. 1998-07-22 Bug: XML names include extra characters.
    XML names may contain `"." | "-" | "_" | ":"' in addition to anything Unicode defines as a letter, combining character, or extender. Colon is the namespace delimiter, and the prefix "xml" is reserved. HTML4.0 has these, too. (1998-07-16)
  40. 1998-07-22 Hole: Parser must call createElement with attr list.
    correctActiveElement is evil. (1998-07-22)
  41. 1998-07-31 Hole: need HookProcessor to run Agent hooks.
    (1998-07-08)
  42. 1998-08-10 Bug: need to give an error on missing handlers.
    (1998-08-04)
  43. 1998-08-11 Impr: tag handler classes need loader methods
    Tagset shouldn't have to know how to load classes for handlers. Having static methods for this would allow tag handlers to manage their own name spaces. (1998-06-16)
  44. 1998-08-11 Hole: New Handlers
    A significant amount of language design is needed here.
    The language design is done and the basics are working; missing handlers are now at the level of bugs.
  45. 1998-08-11 Hole: Loading Tagset from file

    The language design is done and the basics are working; missing handlers are now at the level of bugs.
  46. 1998-08-11 Bug: Tagset files don't have search path.
    Need to check .tss and .tso files. (1998-08-11)
  47. 1998-08-12 Bug: Tagsets don't handle bad .tso files.
    Need to try a separate load so we can fall back on .tss or .ts if the .tso file causes exceptions. (1998-08-11)
  48. 1998-08-12 Bug: Tagsets not loaded from resources. No search path.
    Need to be able to use class loader; check .tss and .tso files. (1998-08-11)
  49. 1998-08-12 Bug: implied -> optional
    Check the semantics of implied in SGML. In any case it's not clear; ``optional'' is better. (1998-08-12)
  50. 1998-08-13 Prop: New InterForm extensions (?).
    (of course, one really obvious one is .dpml -- Document Processing Markup Language. (6-29)
    .xh (mh?)
    ``eXecutable HTML'' (``Meta-HTML'') -- resulting in a MIME type of text/html
    .xx (mx?)
    ``eXecutable SML (``Meta-XML'') -- resulting in a MIME type appropriate for XML.
    .xm (md?) [not done yet]
    ``eXecutable MIME'' (``Meta-Data'') -- the MIME header is returned as part of the document
    (1998-06-18)
  51. 1998-08-13 Hole: Namespaces in Entities, Get, Set.
    Easy thing is to make an entity with a name ending in colon a namespace entity. Look them up as usual. As long as a namespace implements NamedNodeList we're cool (except that in fact AttributeList doesn't). (1998-08-12)
    -- The code is written, but may be flaky. org.risource.dps.Namespace needed for clean implementation.
  52. 1998-08-13 Hole: Must define tags at parse time to cache trees.
    In order to parse correctly, you have to have the whole DTD in advance. Yet another argument for the position that a tagset is a DTD. (1998-06-25)
    -- This is handled by recursive tagsets.
  53. 1998-08-13 ~ Goal: (8-1) Run Most Agents
    Be able to run all of the agents in (my) daily use (i.e. history, etc., but not the wacko stuff in the form server). (1998-06-24)
    -- superceded by the push to revise the language rather than work on the legacy tagset, which may never be completed.
  54. 1998-08-?? Bug: Flag unimplemented legacy options.
    Unimplemented legacy options in <get> etc. have to be brought to the user's attention with error messages. Requires a mode flag or, better, dispatching via the legacy handler first. (1998-06-24)
    -- Note that static caching allows the correct handler class to be loaded without having to hack the search order in loadHandler.
  55. 1998-08-?? Bug: replaceChild when redefining entity or handler
    BasicTagset and BasicEntityTable have to do a replaceChild when replacing a binding. (1998-06-26)
    -- There are problems with sharing.
  56. 1998-08-13 ~ Impr: Character Entities should be defined in tagset
    Should have defined value; copy value into references to short-circuit lookup. TextUtil.expandCharacterEntities can then make use of this. (1998-06-22)
    -- Superceded by passive entities and the need to define entities in tagsets.
  57. 1998-08-13 ~ Impr: Entities need to be initialized by the Tagset.
    The justification for this is that strictly speaking entities are defined in the DTD. (1998-06-18)
    -- Superceded by the need to define entities in tagsets.
  58. 1998-08-13 Hole: Passive [character+] entity class, handler
    Having a separate node class and handler class for character entities would simplify inserting and expanding them. Would break some legacy code, though. (1998-06-19)
    -- Code present but may not be debugged.
  59. 1998-08-13 ~ Wish: toNextNode
    Might want flag in Input's that tells whether the current node has been processed yet. (really relevant only for toNextNode, to keep track of whether we've enumerated the children yet)
    -- probably a bad idea.
  60. Should eliminate the ``7-argument'' action.
    At this point there is nothing using all of its arguments, so it's wasteful. Should have methods on BasicHandler to get attrs, content or content string, etc. Only substHandler and testHandler use cstring. (1998-08-12)
    -- methods on BasicHandler not needed, since they are all in Expand.
  61. 1998-08-25 Bug: Test.trueValue, etc. need context.
    otherwise there is no way to tell whether an entity has a value. (1998-08-25)
  62. 1998-08-27 Bug: Parser.getNextSibling must parse children.
    If current node has unparsed children, the parser must consume them in order to get over to the next sibling. (1998-08-27)
    -- Required adding sawChildren flag to CursorStack.
  63. 1998-09-02 Hole: Sorting unimplemented.
    (1998-08-12)
    -- implemented by Pam
  64. 1998-09-03 Hole: Need a Namespace interface
    This will greatly simplify Index. Also should have a method Context.getNamespace. It would look at not only the entities in each context, but the name of the EntityTable nodes. (1998-08-12)
    -- Some testing needs to be done.
  65. 1998-09-03 Bug: Numeric repeat has problems
    Doesn't use &n; for iteration var; not recognized if stop is only attribute present. (1998-09-03)
  66. 1998-09-03 Bug: Test as condition in if is always true.
    (1998-09-03)
    -- This was introduced when rewriting <if> to use the three-argument action method.
  67. 1998-09-04 Hole: <select> unimplemented.
    Needs to implement everything in XPointer spec, at least. (1998-08-12)
  68. Hole: Namespace changes needed.
    Tagset wants to extend Namespace (sigh). Index -> namespace. Changes in <get>, <set>, etc. (1998-08-05)
  69. 1998-09-09 Bug: XML empty flag not recognized
    At least on <for> elements. Not checking in AbstractHandler.isEmptyElement (1998-09-04)
    -- Needs space to keep from being taken as part of an attribute value.
  70. 1998-09-11 Hole: isChildOf unused in parser.
    (1998-08-12)
  71. 1998-09-18 Check: entity-reference termination in InterForms
    The old parser accepts entity references without a closing semicolon. The new one doesn't. (1998-07-23)
    -- strictEntitiesflag added
  72. 1998-09-21 Bug: Loading tagset from file should use resource.
    ClassLoader.getSystemResource should do the right thing. (1998-08-12)
  73. 1998-10-02 Bug: Problems with literals
    eatUntil broken (both parsers); Output to external form needs not to entity-encode characters.
  74. 1998-10-08 Bug: checkpointing fails: Agency not serializable.
    (1998-10-8)
  75. 1998-10-09 Bug: Agency-xhtml not loaded.
    (1998-10-8)
  76. 1998-10-11 Bug: &attributes:foo; doesn't work as namelist.
    (1998-10-8)
    -- was using instanceof instead of testing asNamelist() != null. Entities test their value for being a namelist.
  77. 1998-10-12 Bug indexed names incorrect in attrs.
    Either they're not getting defined correctly or, more likely, not getting expanded under certain circumstances. The symptom shows up in <a href="&AGENT:foo;">&AGENT:foo;</> (1998-10-11)
    -- Expand.expandEntity: checking for period instead of colon.
  78. 1998-10-12 Hole: PIA datatype transition.
    During the transition Agent and Transaction need to be able to return both SGML and DOM data as needed. Fortunately most of the namespaces contain only strings; it's really only hooks that differ. (1998-06-18)
    -- uses Tabular interface.
  79. 1998-10-12 Bug: Rename TextHandler to avoid case conflict.
    Rename to TextNodeHandler to avoid Windows conflict with textHandler. (1998-08-07)
  80. 1998-11-03 Hole: Legacy Handlers
    cd ../interform/handle; make dps-check for report. (1998-06-15)
    -- officially abandoned.
  81. Hole: Need lookupFile(fn, write) in TopContext.
    (1998-07-08)
  82. 1998-11-03 Wish: ParseTreeIterator
    We will want a ParseTreeIterator that implements both Input and Output (and of course TreeIterator). The main use would be constructing an node and then processing it.
    -- Officially delayed until it reappears in the new DOM.
  83. 1998-11-17 Hole <submit-forms>
    It turns out submit_forms was not defined as a legacy handler. A new tag has been put into the pia-xhtml tagset (in ./tagset/pia-xhtml.ts), and a handler class created in org.risource.dps.handle. (1998-11-09)
  84. 1998-11-17 Hole findInterform with suffixes, forWriting
    Needed for locateSystemResource (1998-11-17)
  85. 1998-11-19 Bug Setting AGENT:xxx to null gives null pointer exc. 
    java.lang.NullPointerException
          at java.util.Hashtable.put(Hashtable.java)
          at org.risource.ds.Table.put(Table.java:107)
          at org.risource.pia.GenericAgent.put(GenericAgent.java:901)
    (1998-11-17)
  86. 1998-11-23 Bug: tsdoc fails when loaded from .tso.
    Attribute lists are missing. Possibly due to entities being hooked up with the wrong namespace, except I can't see a mechanism for that. (1998-10-8)
    -- Turned out to be a bug in serializing ArrayNodeList. GAAK!
  87. 1998-11-19 Hole: Entity definitions.
    <define entity> unimplemented. (1998-09-03)
  88. 1998-11-25 Bug: not propagating verbosity from pia to dps
    (1998-10-9)
  89. 1998-12-04 Hole <put> operation
    <put><name>...<value>... -- this can work with the existing bindings of <name> and <value> (1998-11-02)
    -- -> output
  90. 1998-12-11 Note External Entities.
    org.risource.dps.active.ParseTreeExternal is supposed to correspond to an external entity. <define entity=foo system=url&... should define one. See basic.ts for the relevant attributes: mode, and write-mode. (1998-11-09)
  91. 1998-12-11 Hole <connect> and <status>
    It seems pretty clear at this point that <connect> needs to use an external entity as a ``file handle''. The <status> tag can then return the status of the connection, or of an external entity defined using <define entity=name external=url>. (1998-11-09)
  92. Impr piaUSER, etc should become PIA:user, etc.
    (1998-10-11)
    -- done, but the old names are still around for a while.
  93. 1998-12-14 Bug: Need constructor for stripped tagsets.
    Best done with a predefined tagset (STRIP_ts). (1998-08-07)
  94. Bug GenericAgent.findInterform needs extension list.
    Actually, finding files ought to be in a utility class, because you need to get out: the file pathname, the URL path, and the part of the URL after the file that was found (which gets passed to the interform). (1998-10-13)
  95. 1998-12-16 Bug Problems finding HEADER.html file
    was finding Header.java instead! (1998-12-16)
  96. 1998-12-16 Bug not finding agent data directory in DPS
    (1998-12-16)
    -- This will need more attention later.
  97. 1998-12-15 Impr need tsstrip.ts
    (1998-12-15)
    -- very good, it turns out, for identifying missing end tags in tagset files.
  98. 1998-12-16 Impr move agent-specific tag handlers.
    moved to org.risource.pia.agent directory, with the agent classes. (1998-12-16)
  99. 1998-12-18 Hole: Some <select> subs unimplemented.
    id, xptr, replace; some attributes (e.g. recursive) (1998-09-08)
    -- all but xptr
  100. 1998-12-18 Note: The official package name is org.w3c.dom
    Not clear at this point exactly what use to make of this. (1998-07-16)
    -- imported into CVS tree under src/java/org
  101. 1999-02-02 Bug append element
    <append children> seems able to be able to append text, but not elements. (1999-02-02)
    -- was calling super(true, true) in constructor instead of (true, false).
  102. 1999-02-22 Hole remove legacy tagset
  103. 1999-02-23 Bug HTML tag and attribute names not case-insensitive.
    (1998-02-23)
  104. 1999-02-24 Bug HTML attribute names not case-insensitive.
    (1998-11-17)
  105. 1999-02-24 Bug Passive actors needed
    Need a way of defining tags that self-insert, and are executed purely for side-effects or for an effect on the content. (1999-02-17)
    -- <action mode=XXX> where XXX includes replace-element, replace-content, append-content, silent, delete.
  106. 1999-03-26 Bug Boolean attributes should eval. to name
    ...following the ancient SGML tradition. Best done by fixing what the parser does when no = is present. (1998-11-02)
    -- see BasicParser.java line 214. Unfortunately it breaks some things (like handlers); this needs to be fixed. The problem is that some handlers need to distinguish between an entity which is present but has no value, and one which is present with a non-default value.
  107. 1999-03-2x Bug SGML/XML cleanup
    <p> needs to be non-empty. Boolean attributes need to get their name as value. There needs to be a flag somewhere that says whether to output as SGML or XML. (1999-03-11)
  108. 1999-02-24 Check: attribute and tag case sensitivity
    Nothing is careful about cannonical attribute and tag names yet.
  109. 1999-03-xx Hole: NamedNodeList wrapping a Table
    Want to be able to wrap a Table as a NamedNodeList. This will provide the DPS equivalent of org.risource.sgml.AttrWrap. (1998-06-16)
    -- NamespaceWrap
  110. 1999-04-xx Bug Paths broken
    <output dst='~/foo'> doesn't create foo in the agent's data directory, as it should. <output dst='foo'> does. On the other hand, you need <input src='pia:/~/Admin/foo'> to read it. Basically, agent-relative paths don't work right at all. (1999-03-30)
  111. 1999-04-06 Bug getAttributeString must return internal form
    One might think that everything should return internal form, and ToExternalForm should do the encoding as needed. This would be wrong, though -- it would break writing XML to a stream in a form that can be read back. So toString is right, but getAttributeString shouldn't use it. (1999-03-30)
  112. 1999-04-06 Bug Recursive tags
    Need recursive attribute to <define element> to allow the action to contain the new tag itself. On second thought it would work if we just re-use an existing Handler instead of creating a new one. (1999-01-07)
    -- factorial works, so it must have gotten fixed. If it was ever a bug at all.
  113. 1999-03-xx Bug no way to get name of a node in <select>.
    We can select nodes by name, but not get their names. (1998-10-12)
    -- <name/> (no name specified) does this.
  114. 1999-04-06 Impr the convertToString methods are unused.
    We should either use them, or drop them. (1998-10-11)
    -- dropped
  115. 1999-03-xx Impr: XML / interacts with unquoted attr values.
    Solutions include requiring a space or requiring non-ID values to be quoted. The latter is more correct. Quoting strictness should be a parser parameter for maximum versatility. (1998-09-09)
    -- not really a problem: XML requires quoting; a space is provided on output and is conventional on input.
  116. 1999-04-07 Bug If/elsf/else handlers do not work correctly. elsf clause always returns true. See demo_if.xh. (1998-10-19)
  117. 1999-04-08 Bug: NamespaceWrap doesn't return NodeList for bindings
    (1998-10-11)
  118. 1999-04-08 Hole: Major changes needed for new DOM.
    Do this after Level 1 stabilizes. (1998-08-05)
    --
  119. 1999-04-08 Impr: Add ActiveNode operations to Input, etc.
    Fully supporting ActiveNode as well as Node everywhere in the DPS would give faster, simpler code. It is already impossible to use ordinary DOM Nodes in a Processor. (1998-07-07)
    -- I'm not sure when this was fixed; probably as part of the DOM update.
  120. 1999-04-08 Hole: ParseTreeDocument, DTD, & XML nodes.
    ...are still unimplemented. (1998-06-18)
  121. 1999-04-15 Bug Using Entity instead of EntityRef
    Parser is generating Entity nodes instead of EntityRef nodes for entity references. (1999-04-15)
    -- Along the way fixed the problem of EntityRef having an intrinsic value.
  122. 1999-04-16 Bug Empty entity list is global!
    ... so adding an entity to an (active) element that doesn't have one adds the entity to everything. (1999-04-16)
    -- The culprit was GenericHandler.
  123. 1999-04-16 Bug Extract that modifies structure should preserve list
    All of the structure-modifying sub-elements should simply return the extracted list. (1999-04-16)
  124. 1999-04-27 Bug Entities generated by NamespaceWrap are special.
    Entities generated by NamespaceWrap need to refer back to the original wrapped Tabular object, and re-do the put when their value is changed. (1999-04-16)
  125. 1999-04-30 Bug Need a way to handle attributes of a Namespace.
    When a Namespace is implemented by an Element, it will have attributes. Some of these may, as in Agent, need to be included in the bindings. EntityAttr (or a flag in EntityIndirect) and AttrIndirect? (1999-04-27)
    -- AttrIndirect and some hacks in EntityAttr so the entity namespace can include the attrs.
  126. 1999-05-03 Impr Move get, set to basic.ts.
    Also update basic.html. (1999-05-03)
    -- Move done 5/3.
  127. 1999-05-04 Bug Trim buggy.
    (1999-04-?? pg)
    -- TextUtil.trimListItems was really buggy. Simplified and fixed.
  128. 1999-05-05 Bug Logical handler. A space can change the result. For example: 
    This version returns 1
<logical and><test match="abc">&name1;</test><test match="bbc">&name2;</test></logical>

This version returns 0
<logical and><test match="abc">&name1;</test> <test match="bbc">&name2;</test></logical>
    (1998-10-19 pg)
  129. 1999-05-05 Bug Logical and. Does not return correct (or consistent) values. With two nodes, returns false; with 3 it returns the last item. With 4 nodes (and 4th empty) returns nothing. If a fourth node is given, returns it. Some examples: 
    Returns three
<logical and><em>one</em><b></b><b>three</b></logical>

Returns nothing
<logical and><em>one</em><b></b><b>three</b><em></em></logical>

If 4th node is there, returns "four".
    (1998-10-19 pg)
  130. 1999-05-06 Hole XML Representations of nodes.
    Tagset, Namespace, and org.risource.pia.Agent all need to be representable as XML. Use <ELEMENT> and <ENTITY> tags for the content; construct the objects in the parser. Dump the *.tso files in favor of .tsx. (1999-04-07)
    -- saving works as of 4-27, but could be improved. Only tagsets not done as of 5/6; they're less useful.
  131. 1999-05-07 Impr use buffered readers and writers
    (1999-05-07)
    -- this leads to a significant speed increase: 2x for reading!
  132. 1999-05-10 Bug Buffered writers need to be closed.
    ToWriter invisibly wraps its Writer in a BufferedWriter if necessary. This means that closing the Writer you gave it doesn't necessarily flush the buffer! Probably ought to have Writer.close(). (1999-05-10)
    -- 5/10 checked for close with fgrep
  133. 1999-05-11 Bug submitHandler needs defaults
    Need to supply defaults for method (GET) and URL base. (1999-05-11)
  134. 1999-05-14 Hole: <has-attr> in <extract>.
    (1998-09-08)
  135. 1999-05-14 Bug Then and Else both getting expanded.
    Something about expanding the content, presumably. Handle like logical. Only shows up if there are side-effects. If really needs to be quoted, like repeat. (1999-05-14)
    -- Actually it looks as though this was a mistaken theory, but the handler changes improve efficiency anyway, by making <if> and its sub-elements all quoted.
  136. 1999-05-14 Impr Need <replace node>.
    Want to be able to replace a whole node in its parent. (1999-05-14)
  137. 1999-05-19 Hole Need a way to change tagset in context.
    e.g. when defining an Agent or Tagset: need to switch tagsets for the content. Whether this is done by hacking the parser or interposing a redecorator depends on the current Input's class (i.e. whether it's a ProcessorInput). Should be done in TopContext.subDocument. (1999-04-30)
  138. 1999-05-21 Bug Tagset doesn't get switched in subDocument
    For some reason the parser's tagset doesn't get switched when TopProcessor.subDocument is called with a tagset. (1999-05-21)
  139. 1999-05-28 Bug Permanent tagsets aren't.
    Need to copy the tagset when a document includes a <define>. Otherwise the PIA's permanent tagsets get polluted. We used to have isLocked for this. (1999-03-29)
  140. 1999-06-11 Bug Define entity doesn't eval value.
    (1999-06-11)
    -- Value tag was defined as quoted. Duh.
  141. 1999-06-15 Bug repeat loses text in PIA, not in Filter
    Repeating test &li; yields test abc in the PIA, but test a test b test c from process! (1999-06-15)
    -- It loses all text and empty elements from the top level. Weird. It actually depends on how many elements the repeat is nested inside, and the output goes from a ToWriter to a ToNodeList. The solution was to make the default entity handler a BasicHandler!
  142. 1999-06-16 Bug Active tags not processed at BOF
    If the very first node in a file is active, it is just copied. If it's inside a passive element it gets expanded. (1999-06-15)
    -- BasicProcessor.processNode calls getAction without calling getNode. But getAction depended on setNode to initialize action. The solution is to call getNode at the start of run().
  143. 1999-06-16 Bug Mysterious lossage involving GenericHandler
    Nesting a repeat inside any passive element inside an element whose handler is GenericHandler (even with a definition!) causes empty content to be output only on the first iteration. (1999-06-15)
    -- The problem is CurrentActive.putNode of a node that is already a child of the current node. There was ``excess cleverness'' to prevent a child from being added more than once.
  144. 1999-06-16 Bug Unmatched end tag causes havoc in nodelists.
    The previous node is repeatedly inserted, probably because the end tag isn't getting gobbled. Happens on copy, not process; the bad end tag comment is repeated when quoted. Another depth=2 effect. (1999-06-16)
    -- The problem was the test for an already-present sibling in AbstractParser.toNextSibling.
  145. 1999-06-22 Impr Interface cleanup on Action
    Remove parameters from actionCode, rename to getActionCode. Replace COMPLETED with ACTIVE_NODE. (1999-06-22)
  146. 1999-06-22 Impr Interface cleanup on Cursor
    remove getNode(level) -- unused. Make to* methods return boolean. (1999-06-22)
  147. 1999-06-23 Hole Constructing and editing elements.
    We want to be able to modify the attributes and content of an existing element, or construct a new one out of pieces. Probably by adding back <replace> in <select>, and adding some new ones for value, insertion, etc. (1998-12-15)
  148. 1999-06-24 Hole No implementation for CDATASection.
    (1999-06-24)
  149. 1999-06-24 Hole Comment and string handling
    Need a way to get information (especially entity expansions) into comments and CDATA sections. (1999-05-18)
  150. 1999-06-25 Bug Need <let> for local setting.
    Should be like <set> but operate in the most-local namespace. (1999-06-25)
  151. 1999-07-09 Bug <protect markup> doesn't quote content.
    In other words, it's behaving like <protect result markup>. (1999-07-09) [Reported by Bill]
    -- was dispatching on both attributes.
  152. 1999-07-14 Impr: Interface to push-mode parsers
    Most parsers operate in push mode -- they read their input and call a semantic handler for each ``event''. This requires running the Processor in a separate thread. (1998-08-13)
    -- actually, we just define a new Output called "ToProcessor"
  153. 1999-07-14 Hole Event-driven Processor; SAX output.
    Processor needs to be an Output, and Output in turn needs to be a SAXApplication as well as a tree builder. Processor needs to call element handler's startAction and endAction; on active nodes the former starts building a tree (perhaps by pushing something onto the Output), and the latter interprets the tree. Trivial. Need a flag of some sort that says the Processor is in tree-building mode; event-driven mode is indicated by the lack of an Input. The flag essentially has to be a depth. What we push on the output might be another Processor, or we might do it with a pushInPlace. (1999-06-30)
    -- actually, we just define a new Output called "ToProcessor"
  154. 1999-07-14 Impr Output.startElement parameter change
    Should take tagname and attribute list, not element node. (1999-06-16)
  155. 1999-07-14 Bug Something can't handle a DocumentFragment on output.
    Gives a comment containing the node type. (1999-07-14)
    -- TreeFragment didn't have toString methods. Improved error reporting in TreeNode.
  156. Impr Improved error/message handling
    Create &DOC:errors; entity in process.TopProcessor; Context.message (implemented in util.ContextStack) should copy messages into it. There should also be an entity that says which verbosity level to cut off at -- call it &verbosity; and default it to -1. You'll need active.ActiveNodeList.append(), tree.TreeText.new(String), and Context.getBinding(String name). (1999-06-09)
    -- In progress (Bill)
  157. 1999-08-18 Impr tagsetHandler should use context to load tagsets
    It currently uses dps.tagset.Loader, which precludes using context-specific lookup. (1999-08-18)
  158. 1999-08-18 Impr Tagsets need file of origin, timestamp.
    This permits an up-to-date test. (1999-08-18)
  159. 1999-08-18 Impr Simplify tagset.Loader.
    remove getTagset, require -- both redundant with loadTagset. Add loadTagset(File, TopContext) -- needed for site stuff. (1999-08-18)
  160. 1999-09-16 Bug top.subDocument must take Resource or path
    Otherwise include doesn't work right inside agents. Include is also careless about where it loads tagsets from, and it's unclear whether it should be loading them from the document's home or its own. (1999-09-13)
  161. 1999-10-06 Bug Duplicate property raises an exception
    (1999-10-05)
    -- Turned out to be a consequence of binding an element that already had a parent somewhere else. Oops.
  162. 1999-10-06 Impr: PI comments
    Comments of the form <?-- text ?> should be discarded by the processor rather than being copied. (1998-07-15)
    -- this only needed to be checked for in two places in BasicProcessor.
  163. 1999-10-14 Bug DOCUMENT isn't quoted.
    Contents get processed while reading config, which is always wrong. This may be a deeper problem: tsstrip tries to expand entities! (1999-10-12)
    -- It was indeed deeper. BasicHandler doesn't handle QUOTED content correctly.
  164. 1999-10-18 Impr Rename PROPS -> DOC; DOC ->VAR
    Really want DOC to refer to the properties of the document, just like LOC refers to its location. Use THIS or PROC for the processing-time entities? Find with grep-find for '[^A-Z]DOC[^A-Z_]' (1999-10-14)
  165. 1999-10-28 Bug Index not finding property attrs
    -- Needed getValueNodes(context, name) as well as (name).
  166. 1999-10-29 Bug Define with no content fails
    -- this was considered an error, but in fact it's perfectly legitimate.
  167. 1999-11-01 Bug end tags in PI's generate errors
    They shouldn't be getting parsed at all. It seems to have something to do with encountering entities inside the PI. (1999-11-01)
  168. Impr parser should report implied end tags in XML mode
    (1999-07-09)
  169. 1999-11-02 Bug Non-strict HTML parsing needed.
    Need a "nonstrict" attribute in <tagset>. Distinguish between XHTML (XML) and HTMLX (HTML with eXtensions). Need a more general way to get at tagset attributes. (r2.1)
  170. Impr Control parser strictness in tagset
    It's really annoying to have warnings about HTML stuff. (1999-09-24)
  171. Bug Parser needs error messages.
    Since its a ProcessorInput it can find the Processor's message method. (1999-06-11)
  172. Bug no namespace for all installed agents.
    Best done by making a specialized Entity that knows it's in a NamespaceWrap. (1998-10-12)
  173. 1999-11-04 Impr Output wants close method.
    This would simplify closing streams, writers, etc. when we're done. (1999-05-10)
    -- TreeExternal and ConnectHandler still problematic.
  174. 1999-11-05 Impr Parser should use a LineNumberReader
    Should also have optional filename, both for use in error messages. (1999-07-15)
    -- the filename is no longer needed because you can (usually) get the document name from the TopProcessor.
  175. 1999-11-08 Bug PropertyTable/propertyBuilder buggy
    Doesn't convert to text correctly, and properties not accessible to XML inside. Probably also means that the properties aren't getting copied where necessary. Attributes aren't correctly converted to strings. (1999-11-08)
    -- one problem was using ToNamespace, the other was using <test> as the test property -- <test> is active. (oops)
  176. 1999-11-08 Bug Namespace attrs not copied by handler
    Leaves open the question of whether to build at parse time. (1999-11-08)
    -- Several problems, including incorrect string conversion in AbstractNamespace.
  177. 1999-11-09 Bug Attributes still converting entities
    -- turned out that ActiveAttrList.getAttribute wasn't using Attribute.getNodeValue, as it should have been.
  178. 1999-11-09 Bug Empty list in extract -> bad nesting
    Probably not inputting entire content. (1999-11-09)
  179. 1999-11-09 Bug can't take digits from an entity
    also, handling of "modulus" was apparently totally wrong. (1999-11-09)
    -- turned out to be due to doing too much at parse time.
  180. Impr Tagset inheritance by reference; copy on write.
    In some cases, Tagsets should merely reference their parents, not include them. In particular, this should be done when a non-tagset document contains <define> tags. (1999-04-07)
  181. Bug recursive define entity broken
    Defining an entity in terms of other entities doesn't work in a tagset, even if it's recursive. (1999-07-14)
    -- There are actually two problems here: 1. entities defined in the tagset aren't available when processing it. One can easily argue, however, that this is correct behavior. 2. entities need to be expanded in the value of an entity being expanded.
  182. Impr Simplify Namespace and Context ops.
    getEntityValue, etc. should be removed from Context; the more general operations in Index should be used to give a single point of definition. (1998-10-12)
  183. Bug: Passive Entities needed.
    Passive entities (character entities) and some normal entities have to be defined in the DTD. Passive entities untested. (1998-08-13)
  184. Hole: ToParseTree should take a tagset
    Tagset is needed to attach actions to constructed nodes. (1998-06-22)
    -- probably just needs working shallow copy.
  185. Bug: appending to undefined entity fails.
    <set name=foo><get name=foo>...</set> leaves foo undefined. There must be an explanation... (1998-06-26)
  186. 1999-11-16 Impr base attr for numeric
    Maybe ibase for input base, or else an attribute of <parse> (1999-11-11)
  187. 1999-11-16 Impr Improvements to <weekday>
    Add preferred tagname, <date>. Rename attributes and subelements to make more sense. Add stuff for arbitrary formatting. (1999-11-15)
  188. 1999-12-03 Bug parser doesn't insert </p> properly
    Block tags like <h*> and <ul> have to be notInside <p> for correct handling. (1999-12-01)
    -- Almost certainly still buggy. Really need to look at the DTD and do it right.
  189. 1999-12-10 Bug tagsets not found in CLASSPATH
    because they're at the top level now rather than in the package org.risource.dps.tagset. (1999-12-09)
  190. 1999-12-10 Bug TagsetProcessor not set up with location
    ... and hence can't load parent tagsets relative to document root. (1999-12-09)
  191. 1999-12-14 Bug parent tagsets not loaded by site.
    Not finding a parent tagset in a parent subsite. Even "../tagset" fails. Probably Loader or tagsetHandler is sidestepping the site stuff. (1999-10-04)
  192. 1999-12-15 Bug implicit OR sometimes fails
    Specifically, <get> of two values, the first true, the second false. (1999-12-15) bill
  193. 2000-02-14 Bug Too many tags end paragraphs
    For example, currently <font> cannot enclose multiple paragraphs, because a <p> ends it.
  194. 2000-02-17 Bug if (value==null) in TreeGeneric.setValueNodes
    should have been if (newValue == null) (2000-02-17) J.L. Taylor
  195. Impr Timestamps in tagsets
    Tagsets should have their filename and a timestamp so you can determine whether to reload them. (1999-04-07)
  196. [dropped] Impr force tagset reloading
    Is there a convenient way to force an Agent to reload its tagsets? Checking last-modified dates would also work, but it could get slow. (1998-12-04)
  197. 2000-02-18 Impr index in tsdoc documentation
    -- turned out to be trivial.
  198. 2000-02-22 Bug subst bug
    seems to substitute for one char to the left of the match. (1999-11-23) [barrus]
    -- substHandler just calls the library, so it may be a regexp bug. Possibly related to "$" + the tacked-on terminal null.
  199. 2000-02-22 Bug XML attribute names not case-sensitive.
    (1999-02-24)
  200. 2000-02-22 Bug <date> barfs on null sub-elements
    throws a bogus null pointer exception.
  201. 2000-02-25 Impr Reverse default of case attr
    Make the default be case sensitive, and require case="insensitive" for insensitivity; this makes more sense given that, in the present regexp package, substitution cannot be made case-insensitive. (1999-07-02) [Suggested by Bill]
  202. 2000-02-25 Bug entity/attr values must be in children
    otherwise we are not DOM compliant, and there are some mysterious bugs involving structure-sharing of variables.
  203. 2000-02-25 Bug exception getting empty attributes namespace 
           `

    -- turned out to be TreeNodeList(enum) with null enum.
  204. 2000-03-06 Bug empty active tag with mode=replace-content
    ... expands into an empty start tag followed by untagged content.
    -- The problem turned out to be rather subtle (involving endElement(optional)) and is really fixed only for XML.
  205. 2000-04-14 Impr allow resource with class attr
    configureChild checks for class attribute and loads class by name. Requires configure(...) method on Resource.
  206. 2000-06-07 Hole (WOAD) wrappers, #element, #type in tagsets
    Need to define wrapper element and default bindings for #element, #comment, etc. Default element is handled differently from other node types, which have to be rewrapped as elements by the parser. GenericHandler has to operate differently for non-elements. (2000-05-30)
    -- (element="", documentWrapper, element="#type" 2000-06-05)
  207. 2000-06-26 Bug (woad) parser dies on empty file.
    cgi-bin blows up; CVS doesn't. The difference is that there's a dirIndex for cgi-bin, but it's empty. (2000-06-23)
    -- initial symptom: listing an empty dir. blows up if indexed. The problem was that the dirIndex file was empty.
  208. 2000-08-23 Bug out-of-bounds exception in TreeEntity.startString
    ... when the value starts with a TreeText that has an empty string as its value. The parser doesn't normally generate these, but some computations can, and they may also occur on entities with empty values. (2000-08-23 -- jltaylor@texas.net)
  209. 2000-10-05 Impr passthrough action for elements.
    Not at all clear how best to do this. One possibility is lazy evaluation of &element;. (1998-10-12)
  210. REJ Impr: Filter parse tree input
    Filter should be able to read a serialized parse tree. (1998-06-15)
    -- It should do so if the input file ends in ``.xho'' or ``.xxo''. Rejected because parsing XML is faster!
End of list.
$Id: done.html,v 1.53 2001-01-11 23:37:08 steve Exp $