John Marshall

Admittedly I have never understood why people want to index their FASTQ NGS sequencing reads data in this way! So your explanation that this command is really for use with...

I think trying to fix downstream tools would be an endless task — the 4-line records for FASTQ reads thing is pretty entrenched, and is useful when writing hacky scripts....

I totally agree that the 4-line simplification is particularly useful when writing hacky scripts operating on reads data. James suggests that `fqidx`'s intended data type is FASTQ reference assemblies, akin...

`samtools fastq` deals solely with reads data. As discussed, `fqidx` likely has a dual nature (dealing with two file types represented in the same format) and by default is consistent...

👍 but I think you have them the wrong way around in the first paragraph 😄

Your analysis that this is due to the first file being opened, closed, and then opened again is correct. But a better fix would be to refactor `bam_cat()` and `cram_cat()`...

The body of the `#define` constructs a new type from the argument by sticking a `*` on the end. This is sufficient to construct a pointer type pointing to the...

The contents (a bunch of SAM headers) would be much the same, just the filename would be different. I considered using the existing`.dict` convention for this, but there are several...

1. As described, the intention is for this to be a static file alongside the other bwa index files. Like the `.alt` file, it is prepared separately rather than by...

Superceded by #348, which completely refactors this function.