* XmlTextReader.cs : XmlParserContext was not properly popped on
the expected timing and thus had lost BaseURI, XmlSpace and XmlLang.
svn path=/trunk/mcs/; revision=41021
* XmlSchema.cs : XmlResolver.ResolveUri() may return null.
* XmlParserContext.cs : all the string properties must not be set as
null. Thanks to Joshua.
* XmlTextReader.cs,
DTDObjectModel.cs,
DTDReader.cs : XmlResolver.ResolveUri() may return null.
* XsdValidatingReader.cs : XmlResolver.ResolveUri() may return null.
* XslFunction.cs,
Compiler.cs : XmlResolver.ResolveUri() may return null.
svn path=/trunk/mcs/; revision=38084
* System.Xml.dll.sources : added XmlReaderBinarySupport.cs.
* XmlReaderBinarySupport.cs : added support class for base64/binhex
reading that mostly automates those reader support.
* XmlReader.cs : added CanReadBinaryContent, CanReadValueChunk,
ReadContentAsBase64(), ReadElementContentAsBase64(),
ReadContentAsBinHex(), ReadElementContentAsBinHex() and
ReadValueChunk().
* XmlTextReader.cs : Now ReadBase64() and ReadBinHex() implementations
are moved to XmlReaderBinarySupport. Added CanReadBinaryContent and
CanReadValueChunk overrides. Call Binary.Reset() to enable them.
* XmlTextReader2.cs : added CanReadBinaryContent, CanReadValueChunk.
Added ReadContentAsBase64(), ReadElementContentAsBase64(),
ReadContentAsBinHex(), ReadElementContentAsBinHex() (just because
they are overriden in MS.NET).
* XmlNodeReader.cs : added CanReadBinaryContent and CanReadValueChunk.
Call Binary.Reset() to enable them.
* XPathNavigatorReader.cs : added CanReadBinaryContent and
CanReadValueChunk. Call Binary.Reset() to enable them.
* SubtreeXmlReader.cs, XmlFilterReader.cs :
added CanReadBinaryContent and CanReadValueChunk.
Call Binary.Reset() to enable them.
svn path=/trunk/mcs/; revision=37892
* XmlQualifiedName.cs : in 2.0 it is [Serializable].
* XmlTextReader.cs : removed unused local variable.
* XmlTextWriter.cs : Improved some exception messages.
moved some private fields into #if NET_2_0.
* XmlValidatingReader.cs : rename specifiedResolver to
resolverSpecified and use it in internal get_Resolver().
Added [Obsolete] for ReadTypedValue() whose base is obsoleted.
svn path=/trunk/mcs/; revision=36465
* XmlTextReader.cs : Now attribute normalization is handled on Read(),
not on get_Value. This caused a problem when we use
XmlValidatingReader that value is not always normalized expectedly.
svn path=/trunk/mcs/; revision=36361
* XmlTextReader2.cs : added. It is used to serve public API, plus
entity handling support.
* DTDObjectModel.cs,
DTDValidatingReader.cs,
XmlDocumentType.cs,
XmlParserContext.cs,
XmlTextReader.cs :
Implemented XmlTextReader.EntityHandling and ResolveEntity(). To
implement them, XmlTextReader in 2.0 itself is in XmlTextReader2.cs
and XmlTextReader.cs contains Mono.Xml2.XmlTextReader which is the
same as 1.x XmlTextReader. XmlTextReader2 switches entity reader and
source reader, and never handles tokenization.
The reason for "Mono.Xml2" is to avoid large changes in constructor.f
svn path=/trunk/mcs/; revision=35812
* XmlTextReader.cs : Manually expanded some of AppendValueChar() and
XmlChar.IsWhitespace() which are on critical points on performance
(marked as FIXME, but it depends on JIT).
svn path=/trunk/mcs/; revision=35624
* XmlTextReader.cs : Reduced NameTable.Add() for prefixed names. First,
ReadName() now returns atomized LocalName and Prefix. Second, to
avoid extraneous split, SetProperties() now takes prefix and
localName that might be given from ReadName(). Third, removed
FillNames(). This optimizes "namespace-aware and prefixed document"
parsing 10-20% faster.
svn path=/trunk/mcs/; revision=35622
* XmlTextReader.cs :
With related to AddNamespace(), there was extra string.StartsWith()
calls. Now AttributeTokenInfo.FillNames() is splitted into
FillPrefixAndXmlns() and FillNamespace(), and AddNamespace() is done
at FillPrefixAndXmlns().
svn path=/trunk/mcs/; revision=35601
* XmlTextReader.cs :
Reduces LookupNamespace() call in FillNames(). This improves 6-7%
speedup in some scenario.
Code refactory; Removed unused code. AddAttribute() is used only for
DTD, so renamed to AddDtdAttribute(). For PUBLIC and SYSTEM ids,
FillNames() is not required. Two SetProperties() are ambiguous, so
renamed one (that takes TokenInfo) as SetTokenProperties().
svn path=/trunk/mcs/; revision=35598
* XmlTextReader.cs : LookupNamespace() don't have to be called in
ReadStartTag() to check prefix, since NamespaceURI is filled in
FillNames(). 10% performance up by avoiding this extra check.
svn path=/trunk/mcs/; revision=35590
* DTDReader.cs, DTDValidatingReader.cs, XmlTextReader.cs :
When XmlTextReader is created with NodeType Element or Attribute,
skip possible text declaration (that might be used for resolving
entities) that specifies text encoding. With this fix, bug #61274
must be regarded as fixed.
svn path=/trunk/mcs/; revision=35428
* XmlTextReader.cs : Don't create attribute Value string on Read()
(i.e. create string when get_Value() is invoked.) This optimizes
some kind of XmlTextReader usage significantly such as Skip().
svn path=/trunk/mcs/; revision=35338
* XmlTextReader.cs :
Fixed CheckCharacters setter that always set false.
CheckCharacters check is also required for ReadText(). This fixes
bug #68449.
svn path=/trunk/mcs/; revision=35085
* XmlTextReader.cs : Inside DTD, '<' character inside XML comment
is regatded as invalid. This fixes bug #68410. (Error message was
also not correct.)
svn path=/trunk/mcs/; revision=35045
* XmlTextReader.cs : When Normalization is true, CRLF and CR should
be converted to single LF. This should fix part of bug #62076.
svn path=/trunk/mcs/; revision=33262
* XmlTextReader.cs : Fixed EOF not to return true when it is just
closed. Element and EndElement location is now adjusted to be the
same as MS.NET does. This fixes bug #63505 and #63507.
ResetState() now throws InvalidOperationException() as MS.NET does.
* WriteState.cs : added missing 2_0 enumeration.
* XmlReader.cs : ReadTypedValue() is virtual.
* XmlValidatingReader.cs : ReadTypedValue() is override under 2_0.
svn path=/trunk/mcs/; revision=32558
* XmlTextReader.cs : Fixed several problems on ReadBase64();
Whitespaces should be ignored, and '=' was not skipped correctly.
It caused "unexpected end of document" error at immediate close tag.
svn path=/trunk/mcs/; revision=31209
* XmlAttributeCollection.cs : In .net 2.0, it became sealed class.
* IXmlDataEvidence.cs : not "Evidences" but "Evidence".
* XmlInputStream.cs : new csc rejected implicit int->char cast.
* XmlNamespaceManager.cs : In net_2_0, LookupNamespace(string) and
LookupPrefix(string) does not require argument string as atomized.
Added HasNamespace (string, bool).
* XmlReader.cs : Added Settings and SchemaInfo properties.
* XmlReaderSettings.cs : Added NameTable. Throw XmlException when
attempt to set Schemas.
* XmlTextReader.cs : Fixed "Evidences" to "Evidence". new csc rejected
implicit int->char cast.
* XmlValidatingReader.cs : Fixed "Evidences" to "Evidence".
svn path=/trunk/mcs/; revision=30919
* XmlTextReader.cs : Character range check is always done. It is only
character references for which check is not done when Normalization
is false.
A bit more kind error message for multiple document element error.
* XmlTextWriter.cs : Check if character is valid and entitize when
it was invalid.
svn path=/trunk/mcs/; revision=29426
* DTDObjectModel.cs,
XmlDocument.cs : Close stream always.
* XmlNodeReader.cs : don't expect atomized name in LookupNamespace().
* XmlTextReader.cs : Namespace has was not done. Some tests depends on
this failure and to be fixed.
svn path=/trunk/mcs/; revision=28900
* XmlTextReader.cs : simplified ReadCharsInternal(); use ReadEndTag()
and don't bork against <foo></foo>. Also mofidied this method to skip
<foo/>. Fixed bug #59142.
Modified some comments on the code (FIXME->LAMESPEC et al).
svn path=/trunk/mcs/; revision=28229
* DTDObjectModel.cs,
DTDReader.cs,
XmlDocument.cs,
XmlInputStream.cs,
XmlParserInput.cs,
XmlTextReader.cs,
XmlUrlResolver.cs : removed reference to Mono.Xml.
Made Mono.Xml.* classes internal.
* XmlDocument.cs : output xml declaration for TextWriter output in Save()
* XmlElement.cs : Name character check should be done in .ctor().
* XmlNamespaceManager.cs : Removed dangerous internal .ctor()
svn path=/trunk/mcs/; revision=27149
* DTDObjectModel.cs : (DTDAttributeDefinition) create ArrayList
only when they are required.
* DTDReader.cs,
XmlParserInput.cs : raw tag string now don't have to be kept, while
it cost so much. Removed unused code. 20% to 30% memory got saved.
* XmlNotation.cs : removed unused code.
* XmlTextReader.cs : Removed unused code. ReadString() now uses
XmlReader's internal method. Fixed BaseURI to be conformant with
W3C XML InfoSet and (subsequently) xml:base specification (i.e.
reflect xml:base attribute correctly).
svn path=/trunk/mcs/; revision=25026
* XmlReader.cs : For performance reason, XmlTextReader now uses
base ReadInnerXml() and don't store currentTag array unnecessarily.
* XmlTextReader.cs :
- Fixed GetAttribute(int) that might cause index out of range.
- Now it consumes the text reader by reading blocks, as MS.NET does.
Modified GetRemainder(), PeekChar(), ReadChar() and so on.
- Now it doesn't create Value string for linked nodes, unless it
is actually required.
- some code refactory.
svn path=/trunk/mcs/; revision=23041