feat: Add delta compression support

[itismoej]

Introduce a way to set new Subscription delta option:

SubscriptionOptions subOpts = new SubscriptionOptions();
subOpts.setDelta("fossil")  // <=== HERE
Subscription sub = client.newSubscription("chat:index", subOpts, subListener);

Option delta can only have fossil value now (delta compression based on Fossil algorithm).

When set, client will try to negotiate delta compression in channel with a server. If server allows this delta type – messages in channel MAY be delta-compressed. Whether messages are delta compressed or not depends on server publishing logic.

[FZambia]

Hello @itismoej

Many thanks, this is awesome to get such contributions!

I will review changes very soon, a couple of questions:

1. Just curious, do you work in the same team with @Mahdibn ? Or it's just a coincidence that Implemented Delta Fossil Compression centrifuge-swift#104 is opened almost at the same time ? :)
2. Could you please also add additional test cases from https://github.com/dchest/fossil-delta-js/tree/master/test ? I have not looked closer yet, maybe you already using them, but if not - would be nice to have here, as it's hard to validate the algorithm correctness during the review.

[itismoej]

1. Just curious, do you work in the same team with @Mahdibn ? Or it's just a coincidence that Implemented Delta Fossil Compression centrifuge-swift#104 is opened almost at the same time ? :)
2. Could you please also add additional test cases from https://github.com/dchest/fossil-delta-js/tree/master/test ? I have not looked closer yet, maybe you already using them, but if not - would be nice to have here, as it's hard to validate the algorithm correctness during the review.

Hi Alex,

1. Yes we work together but not on the same team :)
2. Sure. I've already added some test-cases but will add your tests too. Will do this in the PR that I created for the python client too. But, since I have not included the createDelta functionality in the client code as it's not being used, I will not include the tests related to it neither.

[itismoej]

Added test-cases from https://github.com/dchest/fossil-delta-js/tree/master/test

[itismoej]

It's weird that the tests didn't pass!

➜  centrifuge-java git:(master) ./gradlew --version       

------------------------------------------------------------
Gradle 7.5.1
------------------------------------------------------------

Build time:   2022-08-05 21:17:56 UTC
Revision:     d1daa0cbf1a0103000b71484e1dbfe096e095918

Kotlin:       1.6.21
Groovy:       3.0.10
Ant:          Apache Ant(TM) version 1.10.11 compiled on July 10 2021
JVM:          11.0.24 (Eclipse Adoptium 11.0.24+8)
OS:           Linux 6.8.0-45-generic amd64

➜  centrifuge-java git:(master) java --version
openjdk 11.0.24 2024-07-16
OpenJDK Runtime Environment Temurin-11.0.24+8 (build 11.0.24+8)
OpenJDK 64-Bit Server VM Temurin-11.0.24+8 (build 11.0.24+8, mixed mode)
➜  centrifuge-java git:(master) ./gradlew :centrifuge:test

Deprecated Gradle features were used in this build, making it incompatible with Gradle 8.0.

You can use '--warning-mode all' to show the individual deprecation warnings and determine if they come from your own scripts or plugins.

See https://docs.gradle.org/7.5.1/userguide/command_line_interface.html#sec:command_line_warnings

BUILD SUCCESSFUL in 1s
10 actionable tasks: 10 up-to-date
➜  centrifuge-java git:(master)

[FZambia]

Yep, strange. Probably somehow related to current working directory used during test run in CI and the resolution of file paths to load content from.

[itismoej]

Changed to relative path. Maybe that helps!

[FZambia]

One more shot in the dark - maybe return the paths you had and try disabling gradle cache

centrifuge-java/.github/workflows/ci.yml

Line 18 in b2dc056

cache: gradle

?

Unfortunately, I don't understand how testing in Java works under the hood to give a proper advice :(

[FZambia]

try disabling gradle cache

Nope, it won't help.

[FZambia]

@itismoej please pay attention to target files - I see them in tests, but such files are gitignored - and they are missing in PR due to that.

[itismoej]

@itismoej please pay attention to target files - I see them in tests, but such files are gitignored - and they are missing in PR due to that.

Oh! Thanks Alex, I wouldn't notice at all :))

Fixed it.

[FZambia]

I do not see anything else important to tweak here at this point – looks good 👍 I'll now experiment locally a bit, maybe come with some comments after that. Will try to do this till the end of the week.

[FZambia]

Hello @itismoej , here is an update:

• This PR is merged – many thanks, simply an awesome job 👍
• I opened a separate PR - Delta improvements, introducing since option #75 which has some improvements on top on this PR, the most important is that it also handles a scenario when SubscribeReply contains recovered publications (they can also be delta-compressed). It was not obvious because the code was inside Subscription.java and mostly duplicated handlePub method of Client. Now with Delta improvements, introducing since option #75 code will be re-used.
• Also, while testing, found the bug on server side – Fix Fossil delta construction in recovered publications centrifuge#415 – it's rather important to fix, as it could result into bad checksum errors on client side while processing recovered publications. It will be part of next Centrifugo release.

I am planning next to look at centrifugal/centrifuge-python#20 - once it is merged, will proceed with releases of both SDKs - Java and Python. It's better for me because the feature is quite complex and while reviewigng Python lib I can realize sth to be tweaked here. And before SDKs releases Centrifugo with the mentioned fix should be already available.