Updated the max isize to be Integer.MAX_VALUE.

[tfenne]

Description

I'm processing some data with a reference sequence that has long chromosomes, over the BAI limit. I ran into this issue where one (likely chimeric) read-pair has an insert size of over 536,870,912 (aka 1<<29), which causes HTSJDK to throw validation errors.

I looked at the spec and can find nothing supporting a limit on insert size this low. The SAM section defines it as between [−2^31 + 1, 2^31 − 1], and the BAM section dedicates a full int32 t for tlen.

[codecov-io]

Codecov Report

Merging #1410 into master will increase coverage by 0.006%. The diff coverage is n/a.

@@               Coverage Diff               @@
##              master     #1410       +/-   ##
===============================================
+ Coverage     68.102%   68.108%   +0.006%     
- Complexity      8373      8374        +1     
===============================================
  Files            573       573               
  Lines          33965     33965               
  Branches        5668      5668               
===============================================
+ Hits           23131     23133        +2     
+ Misses          8644      8643        -1     
+ Partials        2190      2189        -1

Impacted Files	Coverage Δ	Complexity Δ
src/main/java/htsjdk/samtools/SAMRecord.java	68.323% <ø> (ø)	297 <0> (ø)	:arrow_down:
...htsjdk/samtools/util/nio/DeleteOnExitPathHook.java	90.476% <0%> (+9.524%)	4% <0%> (+1%)	:arrow_up:

[lbergelson]

Seems likely to be an artifact of the bai restrictions, but the history of its creation is lost to git history.

[lbergelson]

Lets see if @yfarjoun can remember any objections too this. I can't think of any though.

[jmarshall]

Historical specification note: this field was initially neglected and only lifted from 2²⁹ to 2³¹ a couple of years after most of the other fields (see samtools/hts-specs@6f8dfe41814df32de75bf16cf2719148c939ae3f). This may go some way to explaining the still-present HTSJDK limit on this field — but it was still lifted almost six years ago!

[lbergelson]

@jmarshall Thank you. Sometimes you just have to give things time, you don't want to rush anything.

[tfenne]

Thanks @jmarshall - I figured it was something like that and I appreciate the archeological help!

[jmarshall]

Tragically I still remember making that change :smile: