- widely adopt the use of ybt abstractions instead of raw dbt management. TODO: the loader and the descriptor are still manually managed.
- remove templates from the perf framework in favor of explicit key size checks that more accurately describe "How Keys/Vals Work", and prevent type related bugs.
- also: removed some dead code, refs #5101



git-svn-id: file:///svn/toku/tokudb@51665 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51651 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51646 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51645 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51644 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51639 c7de825b-a66e-492c-adef-691d508d4ae1

refs #5801 merge 5801 (improve threaded stress test framework) to main: now the loader is used for sufficiently large tables, correctness tests fill tables with zeroes while performance tests fill them with bytes based on compressibility, and both use similar code paths for key/val generation for better consistency and readability


git-svn-id: file:///svn/toku/tokudb@51638 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51529 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51482 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51480 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51477 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51473 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51472 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51471 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51470 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51469 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51468 c7de825b-a66e-492c-adef-691d508d4ae1

Modified loader-stress-test-4z (row count /= 10, row size *= 10) to make it faster in valgrind

git-svn-id: file:///svn/toku/tokudb@51458 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51457 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51446 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51429 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51425 c7de825b-a66e-492c-adef-691d508d4ae1

refs #5672 #5351 increase drd timeouts some more (ugh), make locktree stop reporting false positives


git-svn-id: file:///svn/toku/tokudb@51378 c7de825b-a66e-492c-adef-691d508d4ae1

Fix #5833.  Add a test that notices the lack of locking while initialing a partitioned counter, and fix it.

git-svn-id: file:///svn/toku/tokudb@51376 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51374 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51345 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51327 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51326 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51322 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51310 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51305 c7de825b-a66e-492c-adef-691d508d4ae1

closes #5784 don't run valgrind on tdb-recover because it's statically linked and we can't change --soname-synonyms for just one child


git-svn-id: file:///svn/toku/tokudb@51299 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51266 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51265 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51260 c7de825b-a66e-492c-adef-691d508d4ae1

- minimum key/val size is up to 8 bytes. now perf tests and stress tests have a consistent mechanism for generating keys and values, reducing headache.
- perf tests use a "main" function that generates tables based on the provided key/val size and compressibility, while stress (correctness) tests use a main function that generates tables with a provided (default in all cases, I think) key/val size and fully zero'd values, so the table is in the "correct" zero-sum state. previously this was causing perf tests to generate zer-valued rows always, no matter what row size you wanted or how much compressibility was requested, which clearly skewed performance results (ie: your 100 byte rows are only 8 bytes)
- renamed legacy NULL pointers to nullptr
- moved some "special" operations to the perf/stress tests that use them instead of the global header, to keep things more tidy.



git-svn-id: file:///svn/toku/tokudb@51258 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51256 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51254 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51251 c7de825b-a66e-492c-adef-691d508d4ae1

git-svn-id: file:///svn/toku/tokudb@51240 c7de825b-a66e-492c-adef-691d508d4ae1