Closed Jehops closed 1 year ago
Dear @Jehops,
Thanks so much for reporting. I made some significant code changes to improve computational performance, and this allowed some silly bugs to creep in. These were all "read-faults", so other systems (OS X and CentOS, where I do most of the dev and testing) did not catch them.
Moving forward, I'll make sure to run the entire test suite with the address sanitizer turned out before pushing out new releases.
In the meantime, I patched up and re-released 2.5.46.
Best, Sergei
Thank you kindly @spond! The tests do indeed look good with the 2.5.46hf1
tag.
One nit-picky request. Our port/package system converts tags like 2.5.46hf1
to OS package versions 2.5.46.h1
and those are interpreted to be something of a pre-release to 2.5.46
.
% pkg version -t hyphy-2.5.46.h1 hyphy-2.5.46
<
This means if a user were to have hyphy-2.5.46
installed, pkg upgrade
won't pull in hyphy-2.5.46.h1
. The issue doesn't apply here, because our package is still at hyphy-2.5.43
, and I'll directly upgrade to hyphy-2.5.46.h1
. If there are no constraints on your side, perhaps a tag like 2.5.46.1
would work in these situations.
Dear @Jehops,
Great suggestion. I'll use continued '.' notation in the future since the name of the tag does not matter to me internally. Thanks again for reporting the memory issue.
Best, Sergei
Hello,
When running tests on FreeBSD, there are usually segmentation faults, sometimes for the FEL and/or CONTRAST-FEL tests, sometimes for others tests. Here are two example test runs.