Notes while seeking Hypermail info

The CritMail page

High points on a Google search for “Hypermail”:
Dylan Northrup, HM Dev Center, “backup” of Dev Center. Recent HM on HM. Heavy HM user site with intro material.
A beta version of MIME complient hypermail. Perl version with features — a different style,
How to run HM

I see here that there is a race condition that might explain some transient goofs.

I judge from some archives that Peter’s critmail integration is proceeding.

Here are some assumptions that I would like to verify for a variety of reasons.

The URL that HM attaches to a submission is constant. This makes external references into an archive feasible. I think that I have followed such inward links succesfully.
HM regenerates the four index files ab-initio each time. The commentary seems to suggest incremental processing. I suppose that the only thing that is incremental is assignment of new URLs to new submissions and creation of a new html file for each such.

It occurs to me that it might be feasible to change HM to run in a mode which assumed that it would maintain its virtual memory from one session to the next. I presume that it builds memory structures as it passes over the input file and subsequently builds the index files from these structures. If HM were merely “swapped out” between the production of new index sets, that the production of the memory structures could itself be incremental. Only upon a crash would it “go back to the source”.