Skip to main content

Tangtang Proxy

小黑兔About 2 min文档

Tangtang Proxy

Tangtang Proxy is a tool built into the Heitu software that supports using 【1.14.2 client instead of 1.12.1 client】【2.5.3 client instead of 2.4.3 client】【3.4.3 client instead of 3.3.5 client】to play the game — essentially using modern HD clients as proxies for legacy clients.

Tangtang Proxy is similar to HermesProxy, but differs significantly in programming language and implementation approach.

Advantages Over HermesProxy

  1. The stats panel correctly displays Physical Hit / Spell Hit / Ranged Hit / Ranged Attack Power / Physical Crit / Spell Crit / Healing Power — these cannot be displayed in HermesProxy

  2. Can correctly view the target's enchantments

  3. After being slowed or dying while slowed, the character will no longer be stuck in walking state

  4. Target debuff duration display: casting the same spell on the same target now properly refreshes the duration

  5. Hunter pets will not get stuck in stealth

  6. Undercity - Stranglethorn Vale zeppelin route works correctly

  7. Login queue display works correctly

  8. After a Rogue uses Cold Blood, then Preparation, and then uses a skill that triggers Cold Blood, Cold Blood will properly go on cooldown; also, Preparation will not cause a stuck stealth state

  9. Mage Arcane Missiles correctly display projectile trajectory

  10. Can observe the target Mage's 【Evocation】effect

  11. Can display correct stats for the current version, and uses new equipment models and new item icons wherever possible

  12. When using Data Sync, the backpack will no longer get stuck

  13. After using Warrior Charge / Intercept or other similar abilities, the interface will no longer shake randomly

  14. Completely rewrote Tangtang Proxy's spell queue framework; the new skill architecture no longer requires waiting for the result of the previous spell before attempting to cast the next spell to the server.

  15. Casting off-GCD abilities, such as Rogue's Sprint and Evasion, will no longer have an ~0.3 second delay

  16. When a target with mana is interrupted while casting, the cast bar now disappears correctly

  17. Bulk item auto-selling will not cause disconnection

  18. Fixed the issue of Hunters getting stuck on auto-shot

  19. Fixed the issue where searching secondary categories in the Auction House would auto-select the first tertiary category (e.g., selecting Cloth but only showing Cloth Helms)

  20. Fixed overheal and effective healing logs (absorb-type spells cannot yet be tracked)

  21. Developed a skill queue module to fix stuck skills without adding delay and rewrote the pet skill mechanism, adding pet skills to the skill queue to prevent occasional pet skill sticking

  22. When adding friends or sending whispers, if the addon sends a player name in the format 【Name - Server】, Tangtang will attempt to fix it

  23. Fixed the bug where the raid leader could not assign loot after re-entering an instance or after being promoted to raid leader

  24. Fixed the issue where the character remains in a bow-drawn animation after the target Hunter cancels Auto Shot

  25. Supports multi-boxing on different servers simultaneously

  26. Various bugs that could cause client crashes

  27. And more...

Spell Queue

In HermesProxy or Legacy Clients like 1.12.1, the Spell Casting Flow Is as Follows

Client casts spell --> Server returns result --> Client receives result --> Cast next spell

In the above flow, casting the previous spell requires two network round-trips plus server processing time.

For example:

With 50ms latency, you first try to cast Arcane Explosion but it fails due to GCD. The second attempt succeeds.

In this case, the total time for successfully casting Arcane Explosion is: 50 + 50 + your keyboard input interval + 50 + 50 + two server processing times

Under Tangtang's New Skill Queue Model

Client casts spell --> Server returns result --> Client receives result

Client casts spell --> Server returns result --> Client receives result

Client casts spell --> Server returns result --> Client receives result

...and more

This is now essentially a parallel operation model.

Using the same example:

With 50ms latency, you first try to cast Arcane Explosion but it fails due to GCD. The second attempt succeeds.

At this point, both cast requests hit the server simultaneously:

  1. 50 + 50 + GCD failure
  2. Your keyboard input interval + 50 + 50 + one server processing time

Compared to the traditional model, this saves 50 + 50 + one server processing time — approximately 100 milliseconds