WIP: Feat/windows service

not ready for merging - but ready for review

resolved all threads

There is a lot of magic of installing and running the helper. It looks good to me. I'm wondering if it makes sense to try to factor some of this magic into a module pkg/helper/service or something like that just to don't get overwelmed by it every time we look into the helper code.

But I could be fine merging it in it's current form.

mentioned in issue #222

added 14 commits

4c5fbd10 - 1 commit from branch leap:master
8ee9fb20 - WIP: [feat] windows service
f39a258b - [pkg] add missing icon for demo
712c31cd - [feat] listen on available port
93461eb7 - [refactor] move conditional build instance up
85bf9575 - [ci] fix build errors in dockerfile
f56ed456 - [pkg] change installer to use new windows service
99f89653 - update dockerfile
351943f7 - [pkg] remove nssm from build makefile
a7b9814c - [feat] handle version
1d6ba0ee - [feat] write port in darwin too
6f004650 - [pkg] fix go versions
d2189230 - [pkg] target to build windows pkg in container
e5c6027c - [pkg] update to eoan, add helper script

Compare with previous version

marked as a Work In Progress from kali/bitmask-vpn@8ee9fb20

added 5 commits

567d4925 - [pkg] fail-safe for uninstaller
57ec9f99 - [pkg] another try at removing the nssm helper
732e6871 - [pkg] initializer service early for windows
8f0e759b - [pkg] uninstalling dance for old-style nssm installers
645b2c62 - [feat] probe for the port of our matching helper

Compare with previous version

added 3 commits

62d60d81 - [bug] stop service before uninstalling it
869ef7cc - [bug] delete moar things
e947764e - [bug] clear errors properly on each stop attempt

Compare with previous version

added 21 commits

507fdc4a - [feat] initial implementation of windows service
f611c441 - [pkg] add missing icon for demo
f4f877ca - [feat] listen on available port
2c155ba5 - [refactor] move conditional build instance up
53d08f19 - [ci] fix build errors in dockerfile
04c0b508 - [pkg] change installer to use new windows service
0bc200f1 - [pkg] update dockerfile
9fc35e50 - [pkg] remove nssm from build makefile
6e772ff1 - [feat] handle version
a2da425c - [feat] write port in darwin too
becda788 - [pkg] fix go versions
6fc9af8c - [pkg] target to build windows pkg in container
0d8c9ed9 - [pkg] update to eoan, add helper script
285fb548 - [pkg] fail-safe for uninstaller
5643dea6 - [pkg] another try at removing the nssm helper
f67ef923 - [pkg] initializer service early for windows
52e1ac24 - [pkg] uninstalling dance for old-style nssm installers
5bb56004 - [feat] probe for the port of our matching helper
be1a9f3c - [bug] stop service before uninstalling it
2210ff32 - [bug] delete moar things
5685505e - [bug] clear errors properly on each stop attempt

Compare with previous version

ok - I think this reached a usable form finally. I've uploaded the latest here: https://downloads.leap.se/RiseupVPN/windows/beta/RiseupVPN-0.20.1-21-g869ef7c.exe

I have checked it on a w10 machine, re-installing after 0.20.1 and after itself, and I think it behaves properly (but there could still be some corner cases, like calyx/demolib vpns around...)

@meskio after you can give another quick test I'll update the windows issue so that people can give feedback about this build, and hopefully let's see what AV think of it. I'll get to solve #265 (closed) tomorrow, and then look into building for the store so we could be ready for a release (0.20.4 will basically have this, plus snap/deb rebuild).

Do you think it's wise to send this to the sandbox analysis service at this point? (see #239)

btw I've added some refactor commits recently, mostly to try to make the "black magic" around the windows service easier to follow, let me know if you still think some changes could be made there.

one thing that still annoys me (and I think it might be some kind of go bug or quirk) is that if you cross-compile, you lose the stdout logging/interaction in the helper - at least I think i should document it because it's fucking annoying to debug like this...