Customable weights used in generateRandomizedSpec() #163
Conversation
|
Maybe we should add a field to struct |
|
Thanks for PR and all the help! It looks quite complicated to me and might not be intuitive enough to use. Would you like to add documentation and a few examples? |
u_common.go
Outdated
| // Please make a copy first instead of modifying them directly. | ||
| var DefaultWeights = Weights{ | ||
| Weight_Extensions_Append_ALPN: 0.7, | ||
| Weight_TLSVersMax_Set_VersionTLS13: 0.4, |
There was a problem hiding this comment.
As discussed if we could update this list of weights it would be great. For example, perhaps it is a good time to have TLS 1.3 to weigh more?
There was a problem hiding this comment.
Maybe we should update these preset weights in another PR?
There was a problem hiding this comment.
Sure. I was about to contact Sergey to learn how did he yield all these parameters.
Yes, I feel the same. I will extend struct |
|
Upgraded, and fixed a bug introduced by #155. |
| @@ -2076,11 +2076,11 @@ func (uconn *UConn) ApplyPreset(p *ClientHelloSpec) error { | |||
| } | |||
|
|
|||
| func (uconn *UConn) generateRandomizedSpec() (ClientHelloSpec, error) { | |||
| return generateRandomizedSpec(uconn.ClientHelloID, uconn.serverName, uconn.HandshakeState.Session, uconn.config.NextProtos) | |||
| return generateRandomizedSpec(&uconn.ClientHelloID, uconn.serverName, uconn.HandshakeState.Session, uconn.config.NextProtos) | |||
There was a problem hiding this comment.
| @@ -1847,7 +1847,7 @@ func utlsIdToSpec(id ClientHelloID) (ClientHelloSpec, error) { | |||
| default: | |||
| if id.Client == helloRandomized || id.Client == helloRandomizedALPN || id.Client == helloRandomizedNoALPN { | |||
| // Use empty values as they can be filled later by UConn.ApplyPreset or manually. | |||
| return generateRandomizedSpec(id, "", nil, nil) | |||
| return generateRandomizedSpec(&id, "", nil, nil) | |||
There was a problem hiding this comment.
Actually the original ClientHelloID won't be changed, and it's expected too.
|
If no other thing coming to your mind at the moment let's get this merged. We can always make changes later, before the next tagged version (expected in a few weeks). |
|
It's ready to be merged. |
|
Thanks a ton! Merged. |
I agree. But declaring variables using an exported struct without fields name is a very bad practice and is expected to be broken. As v1.2.1 is a little buggy, it's suggested to tag v1.2.2 ASAP once these preset weights are updated (by you, of course). |
|
Once sergey gets back to me 😉 |
|
@RPRX Actually a covert channel between us would be beneficial given that our areas are largely overlapping with each other. If you wouldn't mind? |
|
@gaukas I've sent an email to you, please check. |
|
I don't think Sergey is responding to inquiries about uTLS. I will get v1.2.2 released as a hotfix to v1.2.1, then we can discuss a smarter way to assign these default weights. |
As described in #157 (comment), allowing users to set their own weights would be helpful (and convenient for some purposes too).
CC @gaukas