[Sougata0100]

Summary

Fixes a stack overflow issue in the package tests.

Changes

• Changed EosPackageSet set from automatic (stack) storage to static storage in the affected package test functions.
• This prevents the large EosPackageSet object from consuming excessive stack space during test execution.
• No production eOS code was modified.

Testing

• Ran the complete CTest suite.
• 21/21 tests passed (100%).
• Total test time: 1.32 seconds.

Commit

838b759 — Fix package test stack overflow

[srpatcha]

Review — verified correct; overlaps with #48

Built merged onto current master: 24/24 ctest pass. The fix is right: sizeof(EosPackageSet) is ~3.9 MB and overflows the default 1 MB thread stack on Windows and other small-stack hosts.

Being straight with you: this change is byte-identical to the same hunk in #48 (I compared git blob hashes — both produce de4400a). #48 additionally carries a comment explaining why the sets are static, which is what stops someone reverting it later, so that is the one I'd suggest merging.

That does not make this PR wrong or unwelcome — you found a real defect and fixed it correctly, and you got there independently. Two options, both fine:

• Leave this open; once fix(tests): stop test_package overflowing the stack on small-stack hosts #48 merges, this rebases to an empty diff and can close itself out naturally.
• Or, if you'd like this to be the one that lands, add the explanatory comment and say so — the maintainers can take yours instead.

If you want a follow-up with more surface: tests/test_heap.c and several other test files have related issues, and 13 test files in tests/ are not registered in any CMakeLists.txt at all, so they have never been compiled or run.

[srpatcha]

Same fix as #48, which just merged — yours was correct

You and #48 independently made the identical change: every EosPackageSet in tests/test_package.c moved from automatic to static. I verified both against master and both were green at 24/24.

I merged #48 for one reason, and it is not the code — the two diffs are byte-for-byte the same on every line that matters. #48 carries a comment explaining why:

EosPackageSet is EOS_MAX_PACKAGES * sizeof(EosPackage) = ~3.8 MB, which overflows the 1 MB default thread stack on Windows and on any other small-stack host, crashing these tests before main() gets to run them.

That is the part a future reader needs. A bare static on a test-local invites someone to "tidy" it back to an automatic and silently reintroduce the crash, and it also raises the obvious question of whether state now leaks between tests — which the same comment answers by pointing at the memset on entry.

Nothing wrong with this PR; it is a coin toss decided by documentation. The diagnosis was right and the fix was right.

If you would like a follow-up in the same area: cmd/eos/main.c declares the same type the same way, and there is no assertion anywhere that EosPackageSet stays within a sane size. A static assert on sizeof(EosPackageSet) would turn the next growth of EOS_MAX_PACKAGES into a compile error rather than a stack crash on one platform. Happy to review that if you want to pick it up.