Rob Patro

> I believe the problem with valgrind and jemalloc concerns finding leaks when the program exits, but that wasn't what it showed here, rather an access violation or use of...

Hi @mathog, Thanks again for all of the detailed information. This has helped us track down the source of these uninitialized reads. We have eliminated them, and valgrind now runs...

Thank you for the detailed new report @mathog! So *this* is behavior we have seen before. On our builds, its sort of rare, but every once in a while, we...

@mathog, If built with just the system malloc, the programs will run fine (i.e. they will produce the correct output). The main reason we link against jemalloc is that both...

@mathog, P.S. As you can see, I reported this upstream to the jemalloc folks. One suggestion they have is to compile jemalloc with `--enable-debug` passed to `configure`. We'll try this...

Yes, the fix is on develop on both. The next salmon release (1.3.0) has a few other improvements and fixes and will likely be out latet this week or next...

Thanks for the patch and the detailed logs. The version of HTSlib is due to the recursive-dependency in [seqlib](https://github.com/walaj/SeqLib). The `all` target actually builds a lot more than just pufferfish....

Hi @Uzaaft, That's a nice idea! While we have some experience with integration and performance regression testing, we've always done that outside of the context of the code tests themselves...

Got it — so I think this would be possible, but would probably require the download of moderately sized external files (stuff that can't easily be put into VC). Presumably...

Hi @wmacnair, Thanks for the detailed question (and for including the nice graphics!). I will also cc @csoneson on my reply for her — always invaluable — take. Regarding your...