Dana Robinson

Keep in mind that this will run veeeeeeeeeeeeeeeeeeeeeeeerrrrrrrrrrrrrrrrrrrrrrryyyyyyyyyyyyyyyyyyyyyy slowly as some of those H5X_DEBUG checks add a LOT of sanity checking. We should probably see how long such a run...

We've been removing much of the bonus debug functionality. We can re-evaluate the necessity of this as we get closer to the 1.14.5 release date.

@opoplawski - Is this still a problem with the hdf5_1_14 branch ?

You can report them to us, but I'm not going to make any promises about debugging issues through 3rd-party software until we've managed to close all of our own oss-fuzz...

@jhendersonHDF , @lrknox - s390x is big-endian. Do we ever see t_mpi failures on our Power system? It's probably too late to investigate this for 1.14.3, but we could build...

HDF4 has historically had a very poor separation of public vs. private interfaces. In 4.3.0, which will release at the end of February, we've finally stopped deploying internal header files...

Unfortunately, there won't be time to address this before the 1.14.4 release. We'll try to get to it over the summer, possibly as a part of larger direct I/O cleanup.

Did you not run CMake to build the files? We don't check in generated files, so unless you run CMake, you will not have all the source code you need...

The problem with that idea is that if it made its way into a release branch, people would be prevented from building the library unless they installed Perl. I'm disinclined...

AOCC works with develop, @lrknox will check to make sure this is true for all maintenance branches