[Mono-devel-list] Re: System.Xml patch

[Atsushi Eno]

Hi Andrew,

Good catch :-) However the fix does not look correct. Your patch
also avoids entity content evaluation, that means it sometimes
results in incorrect well-formedness.

valid-sa-086 looks like this:

	<!DOCTYPE doc [
	<!ELEMENT doc (#PCDATA)>
	<!ENTITY e "">
	<!ENTITY e "<foo>">
	]>
	<doc>&e;</doc>

If there is such xml like:

	<!DOCTYPE doc [
	<!ELEMENT doc (#PCDATA)>
	<!ENTITY e "">
	<!ENTITY e "<foo&>">
	]>
	<doc>&e;</doc>

Then it should be rejected. Note that there is '&' in the second
ENTITY declaration.

So we still have to invoke ResolveInternalEntityReplacementText(),
but without replacing "replacement text". Am not sure I can fix
it soon but would like to take care of it at some stage.

Thanks
Atsushi Eno


Andrew Skiba wrote:
> Hi!
> 
> Please review the attached patch. It fixes one of the regressions in W3C 
> testsuite. I checked it with nunit tests and xsl standalone, it does not 
> break anything.
> 
> The idea of the patch is that "If the same entity is declared more than 
> once, the first declaration encountered is binding" (citation from 
> http://www.w3.org/TR/xml11/#sec-entity-decl ). In this test 
> (valid-sa-086) they redefine an entity with unclosed tag, and current 
> implementation crashes.
> 
> I made the minimal necessary change, but may be you will want to include 
> other flows as well.
> 
> Regards,
> Andrew.