Files
ProcessControl/conv/07-part-VII.md
T
2026-10-06 08:11:00 +02:00

4.5 KiB
Raw Blame History

Conversation Log — Part VII — Now Merge the Two Architectural Ideas · Steps 48–52 (turns 77–83)

⬅ Previous: Part VI · Index · Next: (not started yet) ➡

All parts: Setup & Step 0 · Part I · Part II · Part III · Part IV · Part V · Part VI · Part VII


[77] Student

Syncthing was oversaturated from 23 to 5, so it is normal, concerning the mediator we want t oadd them when there is no direct changes to the data to made, but more like addition (like a duration, adding a type, adding flourish around it, timestamp, ...). Like this we keep the process at what it is intent (rot13 is not intended to give a display color to the text in the console, it is only meant to output the text rot13ed)

Tutor: the case for a mediator, made correctly — the adaptation belongs to neither capability, so it gets its own component. Refined afterwards: the line is drawn by contract compatibility, not by adding versus changing. A Celsius→Fahrenheit mediator changes rather than adds and has earned its place; one that appends a timestamp a consumer already understands has added something and is still suspicious. The question is always whether the consumer could read the producer's message as it stands.

[78] Student

but you have step over 49, i only give answer to 48

[78] Tutor — reply

Correct and my mistake: Step 49's transformer and Step 51's rule were quoted before the student had read them. Apologised, went back to one step at a time, and held Steps 50–52 until he reached them.

[79] Student

i'd say either the same as before (the idea of changing display color in the terminal to show that the output has been rot13ed, or we could imagine an aggregator taking the results of many rot13 before sending them

Tutor: both valid; aggregation is in Step 49's own list. Left him the question of when a batch is complete — count or timeout — as state with an owner.

[80] Student

so it subscribe to rot13 outbox and publish to its own outbox, consoleOut then subscribe to the transformer

Tutor: ✅ and the consequence named: inserting a transformer changes no code. ConsoleOut receives a different Intent (TOPIC display/outbox instead of TOPIC rot13/outbox). ROT13 never learns the transformer exists, the transformer never learns ConsoleOut exists, and each writes only its own outbox.

[81] Student

the idea is to keep each service at its minima, maybe that later we want to change the wiring and add an aggregator after the transformer, so keeping it egoistic keep it easily mutable

Tutor: the flexibility argument, correct. Three further costs of publishing straight into consoleOut/inbox: the transformer would contain ConsoleOut's name (topology inside a service, the Step 28 relapse); that inbox would then have two authorities, payload and Intent; and one inbox serves one consumer, whereas an own outbox serves a logger, a web view or a second display at no cost.

[82] Student

SensorService writes consoleIn/outbox, rot13 writes rot13/outbox, transformer its own

[83] Student

the configurator writes the control inboxes, consoleOut writes nothing, AcuatorService output the aggregateed data

[83] Tutor — reply

Inventory closed, and it passes the one-writer test:

mailbox single writer
consoleIn/outbox SensorService
rot13/outbox ProcessorService
display/outbox the transformer
every control inbox the configurator — the one designated Intent authority
the terminal ActuatorService (not a mailbox, so no authority question)

Part VII complete: egoistic capabilities, Intent from one authority, a transformer only where contracts differ, one writer per resource.

Progress record

  • Current step: Part VII complete → 53 (Part VIII)
  • Last observed result: writer inventory complete, one authority per mailbox
  • Unresolved problem: none
  • Next action: list the filesystem transport's strengths and weaknesses from experience, then compare with Step 53

⬅ Previous: Part VI · Index · Next: (not started yet) ➡

All parts: Setup & Step 0 · Part I · Part II · Part III · Part IV · Part V · Part VI · Part VII