Skip to content

docs: eFlow is genuinely Planned, and the eos pointer was wrong - #1

Open
srpatcha wants to merge 1 commit into
masterfrom
docs/planned-and-correctly-so
Open

docs: eFlow is genuinely Planned, and the eos pointer was wrong#1
srpatcha wants to merge 1 commit into
masterfrom
docs/planned-and-correctly-so

Conversation

@srpatcha

@srpatcha srpatcha commented Sep 1, 2026

Copy link
Copy Markdown
Member

This repository has three files. The README said:

The code that will become eFlow lives in eos
today, at (not yet started; EoStudio is the host).

The parenthetical contradicts the pointer, and the pointer is wrong. eFlow is not
kernel-adjacent and no part of it belongs in eos. §18 of the v2.0 master design
places it inside EoStudio:

EmbeddedOS Studio
  |-- Project / Board Config
  |-- Code / Debug / Trace
  |-- eFlow visual design
  |-- eAI model tools
  `-- EoSim integration

Planned is correct here, unlike its siblings

I checked all three placeholder repositories. eNet and eSec are empty
repositories sitting over working code in eos — their status was wrong and I
have corrected it (eNet#1, eSec#1). eFlow is different: there is no
implementation anywhere in the platform.

$ grep -rl "eflow" --include=*.py --include=*.c --include=*.ts EoStudio/ ebuild/
(nothing)

So Planned stands. What changes is that the README now says why, rather than
implying work is under way somewhere else.

The design constraint worth recording

§18.1: eFlow should generate/configure standard EmbeddedOS components and
manifests; it should not create an incompatible second runtime.

eFlow draws a graph and emits manifests. It does not execute anything. The moment
it grows its own scheduler or IPC, the platform has two runtimes and every
EmbeddedOS guarantee has to be re-established inside the second one.

§18 makes the same point about its host: "The graphical environment is a client
of the platform toolchain, never a second implementation."

Why nothing should be built yet

eFlow's output is component manifests. That format did not exist until
embeddedos-org/embeddedos-stack#20, which is still in review. Building a graph
editor that emits an undefined format means inventing the format in the editor —
precisely how the second runtime §18.1 forbids gets started.

EoStudio is not a clean client of eBuild yet either. It carried a 588-line second
build system with seventeen backends, none of them ebuild, and its one
reference to the tool was a string that is not valid CLI
(embeddedos-org/EoStudio#24, partially addressed by EoStudio#25).

The README now names three conditions, in order, under which this repository
earns code — so "when do we start eFlow" has an answer rather than being
re-argued.

The README pointed at eos as the home of code that will become eFlow, with
the path "(not yet started; EoStudio is the host)". The parenthetical
contradicted the pointer, and the pointer was wrong: eFlow is not
kernel-adjacent and no part of it belongs in eos. §18 places it inside
EoStudio.

Unlike eNet and eSec — whose placeholder repositories sit over working code
in eos — eFlow has no implementation anywhere. Planned is the correct status
and this now says why rather than implying work is under way somewhere else.

Records the design constraint that matters, §18.1: eFlow generates
components and manifests and must not create an incompatible second runtime.
It draws a graph and emits manifests; it does not execute anything.

And states the dependency that blocks it. eFlow emits component manifests;
that format did not exist until embeddedos-stack#20 and is still in review.
Building a graph editor for an undefined format means inventing the format
in the editor, which is how the second runtime §18.1 forbids would start.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant