fix Protocol.js to enable tcp-encryption

[jhgoebbert]

This fixes issues I had with tcp-encryption in combination with the html5 client.
With this MR I can connect from the browser to a xpra server started with:
xpra start --start=xterm --bind-tcp=0.0.0.0:10000 --html=on --tcp-encryption=AES --tcp-encryption-keyfile=auth.txt

• uintToString(..) does not exist. Hence, Uint8ToString(..) from lib/rencode.js imported at line 627 is used
• Utilities is not in scope so Utilities.StringToUint8(..) fails. Hence, StringToUint8(..) got added to class XpraProtocol
• Utilities is not in scope so Utilities.Uint8ToString(..) fails. Hence, Uint8ToString(..) from lib/rencode.js imported at line 627 is used
• ord(..) is not defined. Hence, ord(..) got added to class XpraProtocol

I couldn't figure out why Utilities is out of scope in places where it is needed in Protocol.js - that's a mystery to me.
First I thought it is because of the wrong order Utilities.js and Protocol.js is included in index.html - but fixing this did not help.
(btw: another thing I came across was that Uint8ToString() is defined in Utilities.js and rencode.js the same way ... )

[totaam]

Odd, I remember testing AES mode after fixing lz4 (or was it rencodeplus?).

another thing I came across was that Uint8ToString() is defined in Utilities.js and rencode.js the same way ...

This one also rings a bell. IIRC, copying the function was easier than fixing the import mess.

[jhgoebbert]

I couldn't figure out why Utilities is out of scope in places where it is needed in Protocol.js - that's a mystery to me.
A colleague of mine had a second look at this and comments:
Protocol.js is loaded twice: First with scripts in index.html - (changing the order of the scripts would in principle make the Utilities visible to Protocol.js) BUT that doesn’t matter at all, as what is inside the main if-clause in Protocol.js is not run this first time. That is only called by the line this.worker = new Worker("js/Protocol.js"); (line 40 of Protocol.js) when Protocol.js is loaded the second time. And this time, the scripts that are outside are not loaded as workers run in another global context that is different from the current window.
Therefore the importScripts (on line 615) would need an additional "Utilities.js”.

[totaam]

Therefore the importScripts (on line 615) would need an additional "Utilities.js”.

Wouldn't that be a cleaner fix?

[jhgoebbert]

🤔 probably yes. I will fix that in a new MR