miser

oMiser Technical Note m110400
oMiser Objects (MOb) Directory Operation

orcmid.github.io>

miser> oMiser>


m110400>

0.0.3 2025-11-21T21:21Z

Tentative Overview: Keeping All the Balls in the Air

A Linear Hashing Technique is envisioned to provide access to those oMiser objects that will be locatable via a hash-based search, establishing the MOb Directory for that purpose.

Although every ob-cell has a HashFlag value, that does not necessarily make it possible that an ob-cell can be searched by it’s HashFlag in the MOb. The HashFlag has an often-independent usage solely for optimizing comparisons and, perhaps, reducing occurrence of different ob-cells for exactly the same Ob manifestation.

The critical MOb usage is in the recording of individuals, because generally they are all named. Synthetics are anonymous and in that case they are still closely-coupled into the MOb in order to manage their lifetimes and their connection to defining material.

Since every ob-cell has an applicative interpretation, that is another case for there being associated information and a place where that ties into the MOb for lifecyle management of the ob-cell and associated data as well, akin to synthetics but not individuals. Perhaps having associated material will require being moved out of heap-0 in the first place. This could be very interesting. And maybe fragile as well as challenging.

There is a tie-in to overall oMiser memory management, especially with generational garbage collection. In preparing to scrub a heap-0 by retreating the process stack, it is necessary to identify and move those ob-cells that are still reachable, including from the MOb. So three may need to be an aging process that preserves the ¶-relationship in some manner.

There is also a middle case to consider. Heap-0 ob-cells should not have associated other date (as for synthetics and ob-cells having accerators), because those then have to be found to avoid memory-leaking associated data. Because the whole idea of Heap-0 is to rapidly dispense with short-lived objects, we would like to not have to pay attention to longer-live ob-cells created in heap-0 until they need to be preserved in a deeper generation (or created in a deeper generation in the first place).

So, it would seem, the existence of associated date implies occurrence in the MOb, and there’s some associated lifecycle management principles that allow such objects to be swept through generations so long as they remain accessible, not just being in the MOb. Endurance through generations has to work somehow, ideally without making much material move around. We need an easy way to know what reachable objects live in a given generation so they can be swept out as part of retreating that generation or, alternatively, simply keeping them alive in the generation they occupy.

It’s unclear how one reorganizes or garbage collects generations other than heap-0. It might be that is handled as part of export and reloading. This and other considerations need to be worked out here.

There will also be need for some good simulations and instrumentation in oMiser prototyping to ascertain what the predictable behaviors are and what optimization strategems are generally effective.

Background: How Linear Hashing Came Into the Miser Project Radar

LONG STORY, Details to Follow after all setup here.

Folio Catalog

This placeholder links to raw materials and notes, including text files. There will be organized folios of content as this topic is explored in detail.

ID Status Started Topic
       
m110400a undated 2025-11-16 Notes & Work Items
m110400b 0.0.1 2025-11-21 2025-11-16 The Basic MOb Procedure
m110400b1 preservation 2011-04-21 LwNet Discussion
m110400b2 preservation 1980-10 Litwin1980

I invite discussion about Miser Project topics in the Discussion section. Improvements and removal of defects in this particular documentation can be reported and addressed in the Issues section. There are also relevant projects from time to time.

Hard Hat Area You are navigating the Miser Project on Github

created 2025-11-16 by orcmid