Join GitHub today
GitHub is home to over 50 million developers working together to host and review code, manage projects, and build software together.Sign up
Issue triage for ADAM version 0.20.0 release
Mango needs changes to formats, utils, and adam in order to land their changes from feature branches into master.
Major refactoring around GenomicRDD and related
Updates to Hadoop-BAM to add gzipped/BGZF VCF support, among other things
Improved feature support, including GFF3
BAM header bugs
Support CRAM input/output
Improved VCF support, for Mayo VariantDB Challenge
Support Sample schema
Support Spark version 2.0
Documentation, code cleanup
As I commented over there, I'd need a release with the pipes API (#1114) before December. Is the main goal of pulling the release in to 10/10 to get @ryan-williams off of his #1139 fork? I'm OK with that, but we should really plan to cut a follow-on release (0.20.1?) sooner, because I need #1114, and getting #1114 merged has really dragged out. As of next Wednesday, #1114 will have been open for two months...
My primary concern is probably less about the specific release dates, and more about the fact that they keep on slipping. I'm reasonably OK with putting an arbitrary line in the sand and cutting a release followed up with another release shortly thereafter, but there's a lot of issues here that have dragged out without resolution or action. Specifically, the VCF issues and documentation issues need to close before we can push out the ADAM paper. They're not really negotiable; they need to be done by November. I own two of the documentation issues, and I'll close on them ASAP (#1084, #1085). @heuermh can you commit to a timeline for closing the VCF issues? If not—I hate to say this—but I will take them on and close them. @jpdna can you commit to a timeline for #493?
Of the other open issues, I am OK with triaging #1092 to the "indefinite" pile.
#1114 is nearly ready to go as I see it. I'll work up another external tool example over the weekend to test. Is it ok to block the 0.20.0 release on it being merged?
Would then a Dec 10th target for 0.21 (including #1144 and the follow-on GenotypeAnnotation refactor) be reasonable, with another round of triage to focus on what is necessary for the paper?
#1092 is fairly significant, is it not? No code using the Java APIs will compile without using the non-obvious cast as a workaround.
From today's discussion:
New release date for 0.20 of Friday, 14 October 2016 (@heuermh will cut the release)