Skip to content

vendor: move virtual-sx1262 to 28619507 - #20

Merged
A13xB0 merged 1 commit into
mainfrom
vendor/virtual-sx1262-28619507
Sep 6, 2026
Merged

A13xB0 merged 1 commit into
mainfrom
vendor/virtual-sx1262-28619507

Conversation

@A13xB0

@A13xB0 A13xB0 commented Sep 6, 2026

Copy link
Copy Markdown
Collaborator

Moves vendor/virtual-sx1262 from 0debbb91 to 28619507, the merge of MeshBench/virtual-sx1262#1.

The pin was two commits behind and the consumers check upstream was reporting it on every run:

::warning::MeshBench/meshcore-native pins 0debbb9125bb, 2 commits behind:
           build: the Windows DLL carries its own runtime

What is in the bump

One commit, and it touches one file:

$ gh api repos/MeshBench/virtual-sx1262/compare/0debbb91...28619507 --jq '.files[].filename'
build.sh

Upstream's build.sh now links the Windows shared library statically. Linked the default way it imported libstdc++-6.dll, which exists inside MSYS2 and almost nowhere else, so neither QEMU nor Renode could open the chip model on an ordinary Windows machine and every emulated board there died as a radio that never answered — MeshBench#632.

What it changes here: nothing

Worth being explicit, so nobody reads this as a fix landing.

This repository never had that fault and does not gain anything from its cure. build.sh here compiles the model from source:

vsx="$root/vendor/virtual-sx1262"
if [ ! -f "$vsx/src/VirtualSX1262.cpp" ]; then

rather than loading the shared library the upstream change is about — and for Windows it already links -static -static-libgcc -static-libstdc++ itself. Nothing under src/ moved, so the compiled firmware is byte-for-byte the same decision it was before.

Why take it anyway

A submodule is a commit, and nothing moves it on its own. A pin left behind stays behind indefinitely, and the next upstream change that does touch src/ should arrive on its own rather than stacked behind an unrelated lag nobody has looked at. That is the failure the consumers check was designed to catch, and it caught this one.

🤖 Generated with Claude Code

https://claude.ai/code/session_01RSb8bN9gtL72cXAxtyTbAG

The pin was two commits behind and the consumers check in that repository
was saying so on every run. It now names the merge of
MeshBench/virtual-sx1262#1.

What that commit changes is the upstream build.sh: the Windows shared
library is linked statically, because linked the default way it imported
libstdc++-6.dll and neither emulator could open it on an ordinary Windows
machine (MeshBench#632). Nothing in src/ moved.

So this bump is a no-op for what this repository produces. build.sh here
compiles the model from vendor/virtual-sx1262/src rather than loading the
shared library, and already links -static -static-libgcc -static-libstdc++
for Windows, so the fault never reached the native firmware and the fix
does not change it either.

It is worth taking anyway. A submodule is a commit and nothing moves it on
its own, so a pin left behind stays behind - and the next change upstream
that does touch src/ should arrive on its own rather than stacked behind
an unrelated lag nobody has looked at.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RSb8bN9gtL72cXAxtyTbAG
@A13xB0
A13xB0 merged commit 83006c6 into main Sep 6, 2026
1 check passed
@A13xB0
A13xB0 deleted the vendor/virtual-sx1262-28619507 branch September 6, 2026 20:43
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