On Mon Jun 23 07:30:58 2025 +0000, Nikolay Sivov wrote:
Anything like that looks fragile, if exact compliance with those offsets is necessary. If document structure is simple enough, encoding is UTF-16 always, whole document is available at once, no dtd or entity expansion is necessary, not xsl/xpath is needed, it might even be an option to use some simplified parser similar to what we have for activation contexts in ntdll.
I see, I'll take a look at the ntdll's XML parser. On the other hand, would it be a good idea to add a private wine interface to xmllite that returns the current character position?