Skip to content

Correct the instrument-simulation non-goal in README Scope - #4

Merged
techmanual-ai merged 1 commit into
mainfrom
docs/scope-contradiction-20260916
Sep 17, 2026
Merged

techmanual-ai merged 1 commit into
mainfrom
docs/scope-contradiction-20260916

Conversation

@techmanual-ai

Copy link
Copy Markdown
Owner

What changed

One bullet in the README.md Scope section (README.md:375). Before:

  • No instrument simulation. Tests mock the driver libraries; real use needs
    real hardware.

After:

  • No simulation of your instrument. No driver has a simulation code path;
    lablink-sim is a separate optional bench that the real drivers connect to
    over real protocols, to evaluate LabLink without hardware.

Nothing else in the file is touched.

Why

v0.2.0 shipped lablink-sim — two simulated instruments behind four real protocol front-ends — and the README documents it in a "Try it with no hardware" quickstart roughly 290 lines above the non-goal that denied it existed. The old bullet was simply false, and it was the kind of false that costs a reader trust in the rest of the section.

The replacement keeps the real distinction the section is trying to draw: LabLink will not simulate your instrument, and no driver has a fake code path you might unknowingly be exercising. lablink-sim doesn't violate that — it is a separate optional process the unmodified drivers talk to over real SCPI/REST/serial.

The other non-goal bullets

Checked each against what the repo currently ships; all four are still accurate and were left alone:

  • No server component — nothing hosted or cloud-deployed was added; lablink-sim binds loopback ports only.
  • No GUI — no GUI/TUI code anywhere; lablink-sim is a third console script, not a new interface kind.
  • Not a protocol library — grep over lablink/interfaces/ confirms no driver parses SCPI; the only SCPI parser in the tree is lablink/demo/scpi.py, which is the simulated instrument speaking its own protocol, not LabLink interpreting traffic on the user's behalf.
  • Docker is not a primary install target — no Dockerfile or compose file exists.

The Scope section's intro paragraph (five drivers, external stub, GPIB/RS-485 notes) also still matches the code.

Assumptions

  • Wording is three lines rather than two. Two lines could only fit by dropping either "no driver has a simulation code path" or "to evaluate LabLink without hardware," and both carry weight; three lines matches the existing "Not a protocol library" bullet, so the section's rhythm holds.
  • No CHANGELOG.md entry, per the task — this is a doc correction, not a user-facing behavior change.

Testing

.venv/bin/python -m pytest -q → 350 passed in 13.9s. The suite was green before the change as well; a README edit cannot affect it.

One commit: 2ecc0e9. Not pushed, no PR opened, per instructions.


🤖 Generated with Claude Code

v0.2.0 shipped `lablink-sim`, a simulated bench documented in the README
quickstart, which made the "No instrument simulation" non-goal false a few
hundred lines below the feature contradicting it.

State the distinction that is actually true: LabLink does not simulate your
instrument and no driver carries a simulation code path; `lablink-sim` is a
separate optional bench the real drivers reach over real protocols.

The other non-goals still hold as written — there is no hosted endpoint, no
GUI, no Dockerfile, and no driver parses protocol syntax — so they are left
untouched.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
@techmanual-ai
techmanual-ai merged commit 8de4482 into main Sep 17, 2026
4 checks passed
@techmanual-ai
techmanual-ai deleted the docs/scope-contradiction-20260916 branch September 17, 2026 06:32
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