Repository navigation
Conversation
AmigaOS protection bits belong to the file, not to its contents, and no cross-platform container carries them. A bare `MCPd` ELF downloaded from the release page and moved onto an Amiga over SMB, a USB stick, or a browser arrives with the `e` bit protected; the daemon then refuses to run, and the error looks nothing like the cause. The two installers set the bits themselves, so this only ever bit people doing it by hand -- which is exactly the path a first-time user takes. `scripts/build_release_lha.py` fixes it at the source by assembling the archive *on an AmigaOS filesystem* (a QEMU guest will do; it needs no hardware and leaves nothing behind). It stages the daemon, the five install scripts and a README, sets the executable bit on the binary and the script bit on the scripts, runs `LhA`, and then -- rather than assuming -- extracts the archive again into a clean directory and checks the flags came back before it will hand over the file. The `s` bit is the proof: it is never a filesystem default, so `-s--rwed` on extraction can only have come out of the archive. `MCPd-1.3.lha` (126,535 bytes) is now attached to the v1.3 release alongside the bare ELF, and is what the README and INSTALL.md point a new user at. The release needed nothing else: everything committed after the tag is documentation, and `mcpd/` + `host/` at `v1.3` still match `main`, so the attached binary remains current. Also adds `.gitattributes`. The install scripts are uploaded to targets byte-for-byte from the working tree, and on a Windows checkout autocrlf had been giving every one of them CRLF endings -- a stray carriage return on each AmigaDOS script line, and visible ^M in Amiga text viewers. `eol=lf` pins them, along with the daemon sources the Docker toolchain compiles from the same tree. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01VF4wwBXaKNK2Y2N5sYtt71
Running the installer out of the release archive showed the last line as "keyboard/mouse injection (input.) is DISABLED" -- the asterisk had vanished. In AmigaDOS Echo, '*' escapes the next character, so a literal one has to be written '**'. MCPd-Enable-Input already carried a comment about this; MCPd-Install did not. Found by actually running the shipped artifact rather than reading it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01VF4wwBXaKNK2Y2N5sYtt71
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes the protection-bit problem at its source, so the file a user downloads is correct before it ever reaches an Amiga.
The problem
AmigaOS protection bits belong to the file, not to its contents, and no cross-platform container carries them. A bare
MCPdELF downloaded from the release page and moved onto an Amiga over SMB, a USB stick, or a browser arrives with thee(executable) bit protected. The daemon then refuses to run, and the error looks nothing like the cause.Both installers (
MCPd-Install,scripts/install_mcpd_autostart.py) set the bits themselves, so this only ever bit people doing it by hand — which is exactly what a first-time user does.The fix
scripts/build_release_lha.pyassembles the archive on an AmigaOS filesystem — a QEMU guest will do; it needs no hardware and leaves nothing behind. It stages the daemon, the five install scripts and a README, sets the executable bit on the binary and the script bit on the scripts, runsLhA, and then — rather than assuming — extracts the archive again into a clean directory and checks the flags came back before it will hand over the file.The
sbit is what makes that check meaningful: it is never a filesystem default, so-s--rwedafter extraction can only have come out of the archive.MCPd-1.3.lha(126,574 bytes) is attached to the v1.3 release alongside the bare ELF, and is what the README and INSTALL.md now point a new user at.Verified on a QEMU guest
----rwed(binary) and-s--rwed(scripts)MCPd-Installruns by name, with noExecute— proving the script bit survived — and exits 0Two things found along the way
.gitattributes(new). The install scripts are uploaded to targets byte-for-byte from the working tree, and on a Windows checkout autocrlf had been giving every one of them CRLF endings — a stray carriage return on each AmigaDOS script line, and visible^Min Amiga text viewers.eol=lfpins them, along with the daemon sources the Docker toolchain compiles from the same tree.MCPd-Installprinted(input.)instead of(input.*).*is the AmigaDOS Echo escape character; a literal one needs**. Found by running the shipped artifact rather than reading it.Release status
The v1.3 release needed nothing else. Everything committed after the tag is documentation, and
mcpd/+host/atv1.3still matchmain, so the attached binary remains current.🤖 Generated with Claude Code
https://claude.ai/code/session_01VF4wwBXaKNK2Y2N5sYtt71