Maybe add ShadowsocksR support?

[peter17ji]

ShadowsocksR provides far better protection than the original Shadowsocks. It's using better encryption, obfuscation, and can pretend to be browsing another website. And the R version is technically backward compatible.There's someone already made one using electron framework. https://github.com/erguotou520/electron-ssr

[trevj]

@peter17ji Thanks for the suggestion! I agree at some point we may want to use a different library and/or protocol but I'm afraid we have no plans to do that right now.

[66beta]

Excuse for my poor english.

I bought a service that only provide SSR protocol, And I'd like to use OUTLINE rather than SSRR client.

Consider support this feature when @peter17ji reach 50 thumbs?

[peter17ji]

@66beta Using SS protocol on a SSR-only server could lead to a GFW'd and unusable server, so consider other clients.

[patrickmau]

I live in China and spent quite a bit of time setting up and testing both Outline and SSR servers in Japan, Korea and Singapore. Some observations below:

• Outline (SS protocol) does not achieve decent speeds in China and maxes out at 12mb/s for China Telecom users
• SSR servers using Shadowrocket (iOS) and ShadowSocks client for MAC reach up to 29mb/s
• Outline (SS) does not work at all with China Unicom (reaching a max DL of 5mb/s) while China Mobile works only a little faster, maxing at 8mb
• SSR on iOS with China Unicom does not exceed 2mb/s, making me believe that China Unicom is already throttling SSR connections.

Note: all servers were speed optimized with BBR installed.

My opinion is that if Outline is really to succeed in China, then it would have to adopt SSR standards some time in the future. SS is already getting blocked and throttled here and I dont see it surviving for long anymore, hence would very much support SSR support in the future.

[intellix]

@patrickmau do you have more details about the time of day and which servers you we're trying for those speed tests? For me it's entirely based on when the tests are run.

In the morning @ 9:38am from Shanghai on China Unicom 4G using "Speedtest by Ookla" with an Outline server setup in Google Cloud asia-east2-a I'm getting results like:

Tai Po (Hong Kong) - 26.7Mbps DOWN - 3.70Mbps UP - 28ms
Hutchinson, KS (USA) - 24.9Mbps DOWN - 3.64Mbps UP - 12ms
Derry (Ireland) - 9.09Mbps DOWN - 4.88Mbps UP - 332ms
London (UK) - 5.6Mbps DOWN - 4.45Mbps UP - 323ms
Madliena (Malta) - 8.01Mbps DOWN - 4.78Mbps UP - 359ms

Anecdotally, if I ran the same results tonight between 6pm -> 1am when the Great Firewall kicks in, it'd outright not work or I'd get throttled to hell to the point of un-usability

[fortuna]

SSR does not use better encryption. That's misleading and we should stop spreading this kind of disinformation. SSR only adds obfuscation.

It's very easy to detect SSR. It randomizes the packet sizes, so all you need to do it look at the entropy of the packet sizes. You don't even need to look at the content. Max Lv has implemented a proof of concept.

Because it's easy to detect, we don't intend to support it.