Benjamin Niemann wrote:
If you write a HTML parser, do not implement size limitations. If you are
authoring HTML code, you'll win a prize, if you manage to exceed the 64k
limit with meaningful data...
Been there, done that.
For metadata embedding, it's (reasonably) common practice to embed
annotations and descriptions as attributes that are lifted from a
Dublin Core (or similar) "pseudo-namespace". These aren't part of valid
HTML, but XML namespacing isn't really available to the web yet and
it's long-established web practice that extra attributes are silently
and safely ignored by user-agents that don't understand them.
Once you're adding annotation or metadata, it's not hard to start
breaking 64k length limits, if you're unaware of the issue.