[kiwidude]

@chaley - that was my original intent of ignoring the linux users who have a disproportionate amount of support required.

Then someone submitted a PR for this so I thought push it out give it a go. Except it turned out they never tested their fix. Then they submitted another PR - I pushed that out, didnt fix the issue either, once again they hadn't tested it.

So then I tried myself just abandoning those changes and taking on the latest python3 version of httplib2. Except internally they made so many refactorings to that with dependencies on additional lib modules (pyparsing) and imports that it won't work as a calibre plugin without further hacks and tweaks.

A second person just now has submitted another PR - but their PR makes no sense at all and when questioned they said their change they tested was with a different version of the codebase of the plugin. I've given them the opportunity to send me the files they have actually tested to see if something can be resurrected from that as the last roll of the dice.

When Kovid moves to 3.12 then this is all going to have to get sorted out as presumably this key_file stuff will impact windows users. So I was willing to do a little effort if it solved the future problem. But people submitting untested PR's has really pissed me off. This is one situation where "hey at least they had a go" is not true and commendable - I've wasted at least 8-10 hours on this nonsense as a result...