@ic3l9@newt well, indeed XHTML ≠ XML, though the former may be viewer as the latter, forgetting about meaning of the tags, and indeed it is parsed as XML first and then the entities get interpreted
(I kinda know the difference, I generated #XHTML from custom #XML with #XSLT and wrote custom #DTD for the validation of the XML)
unlike "HTML5" e.g. all the tags have to be closed or the browser will just signal an error and won't show anything — that's a good thing, and the reasons are the same as for strongly, strictly, and statically typed GPPLs
problem with #JSON is not that it's a bad format (although it really is) but that it isn't human read-/writeable and discourages separation of internal data-structures from protocols — data-structures may vary between implementations and versions but protocols must be stable, and just passing lists of hash tables as they are is not really a protocol design
of course all windows behavior is intentional, even if malicious — never mistake hostility for stupidity
specifically in case of #Thunderbird, things like login failure for one server of one account block everything else, and the dialogs can't even be closed oftentimes — if it's intentional then they're just evil or it's a sabotage, should be avoided in either case