[Mono-devel-list] Patch for full-featured mcs /doc support

[Atsushi Eno]

Ok, the harness.mk problem was from both xmldocdiff.cs and
makefiles. Additionally I have to modify decl.cs a bit (to
mimick csc output for invalid "include" element) and had
to make a trick for xml-025.cs that handles inclusion that
csc cannot handle '/' file path separator expectedly.

Sorry for flooding the patches. Please review this latest one.

Atsushi Eno

Atsushi Eno wrote:
> Hello,
> 
> Thanks for several checks!
> 
> 
>>+ public string consume_doc_comment ()
>>+ {
>>+ if (xml_comment_buffer.Length > 0) {
>>+ string ret = null;
>>+ if (xmlCommentSavePoint > 0) {
>>+ ret = xml_comment_buffer.ToString (0, xmlCommentSavePoint);
>>+ xml_comment_buffer.Remove (0, xmlCommentSavePoint);
>>+ } else {
>>+ ret = xml_comment_buffer.ToString ();
>>+ xml_comment_buffer.Length = 0;
>>+ }
>>+ xmlCommentSavePoint = 0;
>>+ return ret;
>>+ }
>>+ else
>>+ return null;
>>+ }
>>
>>The last else is useless and xmlCommentSavePoint = 0 should be only for
>>xmlCommentSavePoint > 0.
> 
> 
> Ok. Actually I noticed that push_xml_comment() is not required anymore
> (since I implemented invalid comment state transition), I just removed
> it as well as xmlCommentSavePoint which is also not required anymore.
> 
> 
>>+ public enum XmlCommentState {
>>+ OK,
>>+ NG,
>>+ Error
>>+ }
>>
>>What is NG ?
> 
> 
> OK, I found that NG is kind of Japanglish ;-) so just renamed
> OK as Allowed and NG as NotAllowed.
> 
> 
>>+ MemberInfo mi = FindDocumentedMember (type, memberName,
>>parameterTypes, ds, out warnResult);
>>+ switch (warnResult) {
>>+ case 1581:
>>+ Report.Warning (1581, 1, Location, "Invalid return type in XML comment
>>cref attribute '{0}'", cref);
>>+ return;
>>+ case 1584:
>>+ Report.Warning (1020, 1, Location, "Overloadable {0} operator is
>>expected", parameterTypes.Length == 2 ? "binary" : "unary");
>>+ Report.Warning (1584, 1, Location, "XML comment on '{0}' has
>>syntactically incorrect attribute '{1}'", GetSignatureForError (), cref);
>>+ return;
>>+ }
>>
>>Why don't do it inside FindDocumentMember to avoid copy&paste.
> 
> 
> Fixed.
> 
> 
>>- : IDENTIFIER
>>+ : IDENTIFIER
>>+ {
>>+ tmpComment = Lexer.consume_doc_comment ();
>>+ Lexer.doc_state = XmlCommentState.OK;
>>+ }
>>OPEN_PARENS opt_formal_parameter_list CLOSE_PARENS
>>{
>>oob_stack.Push (lexer.Location);
>>
>>- current_local_parameters = (Parameters) $3;
>>+ current_local_parameters = (Parameters) $4;
>>}
>>opt_constructor_initializer
>>{
>>Location l = (Location) oob_stack.Pop ();
>>- $$ = new Constructor (current_class, (string) $1, 0, (Parameters) $3,
>>- (ConstructorInitializer) $6, l);
>>+ $$ = new Constructor (current_class, (string) $1, 0, (Parameters) $4,
>>+ (ConstructorInitializer) $7, l);
>>}
>>
>>I am confuse, what has been changed.
> 
> 
> I cannot understand what you mean. It should not be regarded as
> confusion, since it must happen whenever any of you guys have to
> insert logic between tokens.
> 
> 
>>+ } catch (NotImplementedException ex) {
>>+ Report.Error (1569, "Error generating XML documentation file '{0}'
>>('{1}')", xml_documentation_file, ex.Message);
>>+ return false;
>>
>>Does make a sense to catch this exception ?
> 
> 
> Oops, it should just be Exception (to catch xml-output errors).
> 
> 
>>+ vd.DocComment = ConsumeStoredComment ();
>>+ Lexer.doc_state = XmlCommentState.OK;
>>
>>Why don't call this tricky property in ConsumeStroredComment I found
>>only one place where this combination is missing (bug ??). Should be
>>possible also to rename this property ?
> 
> 
> Made as such.
> 
> 
>>BTW: I think we should call this call only for /doc option.
> 
> 
> Ok, now everything in .jay checks RootContext.NeedDocument.
> 
> I've attached the latest diff in mcs/mcs and mcs/tests, and added
> new tests in mcs/errors (cs1587-*.cs). It still needs some love
> on the files in tests (test-harness is not working properly; am
> asking Hari about that).
> 
> Atsushi Eno

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: mcs-doc-20041126.diff
Url: http://lists.ximian.com/pipermail/mono-devel-list/attachments/20041126/fdad09f0/attachment.pl 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: mcs-doc-tests-20041126.diff
Url: http://lists.ximian.com/pipermail/mono-devel-list/attachments/20041126/fdad09f0/attachment-0001.pl 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: xml-025-ref.xml
Type: text/xml
Size: 730 bytes
Desc: not available
Url : http://lists.ximian.com/pipermail/mono-devel-list/attachments/20041126/fdad09f0/attachment.xml 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: xml-025.cs
Url: http://lists.ximian.com/pipermail/mono-devel-list/attachments/20041126/fdad09f0/attachment-0002.pl 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: xmldocdiff.cs
Url: http://lists.ximian.com/pipermail/mono-devel-list/attachments/20041126/fdad09f0/attachment-0003.pl