Closed bertcarnell closed 4 years ago
two issues described in this valgrind trace
1) Invalid read of size 4 in
oacpp::oaconstruct::bosebushl(oacpp::GF&, int, bclib::matrix
I have isolated the invalid read in boasebushl https://github.com/bertcarnell/lhs/commit/6c049248e9fdf49ab35970e37f63d5cc70303155
I also saw a way for the RNG seed pointer to be lost if the geneticLHS threw an exception. I added try-catch blocks to free the memory: https://github.com/bertcarnell/lhs/commit/1a5152e4e9cfc14d927fb712e9d314ff002d8f5f
There still seems to be a valgrind problem remaining, however.
Fixed with this commit: https://github.com/bertcarnell/lhs/commit/3e9d73ae11d6078874073cc802e6dfe5b3866d70
Memory leak not found in R CMD check --use-valgrind.
https://cran.r-project.org/web/checks/check_results_lhs.html
https://www.stats.ox.ac.uk/pub/bdr/memtests/valgrind/lhs/tests/testthat.Rout