Hashtab

[zuiderkwast]

This is in early stages.

Implementation and documentation is in src/hashtab.c and src/hashtab.h.

To test, look at src/unit/test_hashtab.c

To run test, do

cd src
make valkey-unit-tests
./valkey-unit-tests --single test_hashtab.c

TODO

• Very few unit tests
• Basic unit tests of the implemented functionality
• More tests, large tables, fuzzing
• Debug crash or hang for large tables (> 3M elements)
• Type (struct with callbacks)
• Create and release, seed hash function, etc.
• Mem usage
• Global resize policy (allow/avoid/forbid)
• Incremental rehashing, rehash step, expand-if-needed, etc.
• RehashMicroseconds
• RehashingInfo
• Basic operations: find, add, replace, delete
• FindEntryByPtrAndHash (needed?)
• Scan
• Possibility to defrag (flag to scan to emit pointer to the elements within the table)
• Iterator
• Random element
• Fair random element
• Two-phase delete (TwoPhaseUnlinkFind/TwoPhaseUnlinkFree)
• Two-phase insert (FindPositionForInsert/InsertAtPosition)
• Stats

Comparison with dict

Most of the API is similar. The main difference is that the hashtab contains
elements (where an element can contain a key and other data), rather than keys
and values.

dict	hashtab	Notes
key-value entry	element	element contains key
DICT_OK, DICT_ERR	1, 0
dictResizeEnabled	hashtabResizePolicy	Better name
dictAddRaw	hashtabAddOrFind	Better name
dictUnlink	hashsetPop	Delete and return element
dictPauseAutoResize	hashtabPauseAutoShrink	Affects only shrinking, not expanding.
dictResumeAutoResize	hashtabResumeAutoShrink	Does a shrink if needed. The dict one doesn't.

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 347 lines in your changes missing coverage. Please review.

Project coverage is 69.88%. Comparing base (6172907) to head (52e8bba).

Files	Patch %	Lines
src/hashtab.c	0.00%	347 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff              @@
##           unstable     #954      +/-   ##
============================================
- Coverage     70.30%   69.88%   -0.43%     
============================================
  Files           114      115       +1     
  Lines         61744    62091     +347     
============================================
- Hits          43412    43392      -20     
- Misses        18332    18699     +367     

Files	Coverage Δ
src/hashtab.c	0.00% <0.00%> (ø)

... and 11 files with indirect coverage changes

[zuiderkwast]

Moving this to personal repo and hiding from public PRs, for now. It's not ready for public review.