Parser interprets empty read as end-of-file

[rob-p]

Hi Guillaume,

@roryk noticed this issue in RapMap (COMBINE-lab/RapMap#19), and has a nice sample dataset to reproduce it. Basically, when a read is empty, but not the actual end of the file, the parser thinks it's done. Is there an easy way to change the parser logic to account for this situation?

--Rob

[gmarcais]

Indeed, that is not good. I'll fix it ASAP.

[gmarcais]

Can you try the develop branch? The issue should be fixed.

[rob-p]

That was fast! I'll test it out as soon as I get into work.

[rob-p]

Here I betray my strong CMake bias — what magical invocations must I make to generate a proper configure script? I tried aclocal && autoconf && automake --add-missing.

[gmarcais]

The magic call is:

echo abracadabra; autoreconf -fi

[rob-p]

ahhh, see I was missing the abracadabra incantation! Any interest in building a tagged tarball (i.e. with a stable GitHub url) that already has the ./configure script ready to go? Right now I'm pulling from https://github.com/gmarcais/Jellyfish/archive/develop.zip , which, I imagine, is subject to change in content ;P.

[gmarcais]

I will do that. I was hoping to get a feed back from you that it indeed fixed the problem before generating a new version and upload it to the releases.

[rob-p]

I can confirm that this does, indeed, fix the problem (using Rory's test data). The parser now returns (and RapMap now reports) having observed 5 reads, instead of 0, which is correct.

[gmarcais]

OK, the new distribution tarball is ready in the releases(https://github.com/gmarcais/Jellyfish/releases).

Thanks for the bug report.

[rob-p]

Fantastic. Thanks, Guillaume!

[roryk]

Thanks!

[mr-c]

Is it known how far back this bug has existed?