Lua: a missing hashtable string is the null string

[Frotty]

On Lua a missing hashtable string did not compare equal to null, as it does on Jass:

let h = InitHashtable()
if LoadStr(h, 1, 1) == null   // true on Jass and in the interpreter, false on Lua

On Lua, Wurst writes the null string as "": EliminateLocalTypes turns the null literal into "", and a native's nil guard answers "" for a string (LuaNativeLowering). The hashtable emulation was the exception. Both its in-place load and its __wurst_LoadStr helper answered nil for a missing string, so the comparison above became nil == "". A string passed on to a native was nil too. Only a concatenation was safe, because it guards a call's result with or "".

What changes

• A string load answers "" when nothing is stored, both in place ((h.__wurst_ht_str[p] or __wurst_htEmpty)[c] or "") and in the helper. Handle loads still answer nil, which is the null handle.
• LuaNativeLowering.neverNil knows that a string load of the emulation is never nil, so a concatenation of one adds no second or "". A concatenation prints the same as before.

A native declared without the BJ flag (an own @extern native LoadStr) still calls the game's native where the game has one; this changes only the emulation.

Tests

• hashtableOperationsBehaveAsTheNatives (Jass, interpreter and Lua) and its optimised variant now also check LoadStr(...) == null under an absent parent key and an absent child key. anOperandWithAnEffectKeepsTheCallOfTheHelper checks the same through the helper. All three fail on master on Lua only.
• hashtableOperationsArePrintedWhereTheyAreCalled checks the in-place shape and that a concatenation adds no guard of its own. hashtableLoadsUseTypeSpecificBuckets pinned the old nil and now pins "" for strings and nil for handles.
• Passed: LuaHashtableTests, LuaNativesTests, LuaTranslationTests, LuaBackendAuditTests, StringByteSemanticsTests, CompiletimeHashtableTests. The full suite is left to CI. Not run in game.

[Frotty]

@codex review

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-11T08:02:52.234852Z	266497b	Manual request

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. What shall we delve into next?

Reviewed commit: 266497b8b2

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[Frotty]

Closing. Measured on the 3.0.0 client, Jass keeps the null string and "" apart: null == "" is false, and a missing LoadStr equals null but not "". Making a missing string read as "" on Lua trades the == null mismatch for an == "" one. Matching Jass needs the null string to be nil on Lua, with nil-safe string natives where the game's Lua rejects nil (StringLength(nil) raises in the game, Jass answers 0). That is a separate change.