Vision for pysam2

[kevinjacobs-progenity]

Kevin's vision for pysam2

0. Require Python 3 (perhaps 3.7+)
1. Upgrade to htslib 2
   • Many many API updates
   • Ensure all pysam code is 64bit clean to match new htslib API
   • Adopt new region parser
   • Avoid need to vendor htslib
   • Work with htslib team to remove few remaining abort()s in the library code
   • Work with htslib team to provide noninvasive output/warning/error hooks
   • Integrate htslib logging with Python logging
2. Provide a cleaner and more consistent API
   • Complete deprecation of older APIs
   • Harmonize various HtsFile subclasses
   • Deprecate legacy VCF parser
3. Support new VCF features
   • Follow progress in hts-specs (VCF 4.4), htslib
   • URL encoded info and format values (needed to escape VCF delimiters in values)
4. Improve Python developer/user experience
   • Prevent virtually all crashes/segfaults from Python wrapper
   • Fully control C struct lifetimes (we're close, I think)
   • Hide all memory management details
   • Support efficient header translation without user intervention to hide
     translation details from one file/header to another. Currently users
     need to know about TIDs, etc.)
   • Prevent or catch and recover from illegal operations
5. Provide on high quality comprehensive documentation -- a historical weakness
   • Poor/incomplete documentation has been the biggest weakness of pysam
     historically.
6. Modernize test suite
   • modernize current test suite
   • add more high code coverage unit tests
   • add lightweight system tests
7. Simplify scope by wrapping only htslib, not samtools or bcftools
   • Subject to debate of course. I don't use this functionality, so it
     isn't important to me. This becomes a much easier feature to support
     if a developer who cares greatly about this feature steps up to work on
     this.