I sat down and had a look at this... The comment fetching code is based heavily on the old ljdump code, and it appears to "catch up" with un-fetched comments all at once, no matter how many comments that could potentially be.

So, on the initial run, it will attempt to fetch the ID of every comment in the journal, and then attempt to fetch the body of every comment it gets an ID for, which could potentially be many thousands.

To allow it to fetch comments in intervals I'll need to do some re-wiring. Bear with me...