Changes between Version 1 and Version 2 of Ticket #69836

Timestamp:

Apr 26, 2024, 10:28:34 PM (3 weeks ago)

Author:

ryandesign (Ryan Carsten Schmidt)

Comment:

I updated kicad to 7.0.11 on April 7 and the checksums were correct at that time. I agree that they are incorrect now:

% sudo port fetch --no-mirrors kicad
Portfile for kicad changed since last build; discarding previous state.
--->  Fetching distfiles for kicad
--->  Attempting to fetch kicad-7.0.11.tar.bz2 from https://gitlab.com/kicad/code/kicad/-/archive/7.0.11
% sudo port checksum kicad          
--->  Verifying checksums for kicad
Error: Checksum (rmd160) mismatch for kicad-7.0.11.tar.bz2
Error: Checksum (sha256) mismatch for kicad-7.0.11.tar.bz2
Error: Checksum (size) mismatch for kicad-7.0.11.tar.bz2
Error: Failed to checksum kicad: Unable to verify file checksums
Error: See /opt/local/var/macports/logs/_opt_local_var_macports_sources_rsync.macports.org_macports_release_tarballs_ports_science_kicad/kicad/main.log for details.
Error: Follow https://guide.macports.org/#project.tickets if you believe there is a bug.
Error: Processing of port kicad failed

The difference between the file we mirrored April 7 and the file available from GitLab today is:

• kicad-7.0.11/thirdparty/argparse/tools/build.bat

  diff -ru macports/kicad-7.0.11/thirdparty/argparse/tools/build.bat gitlab/kicad-7.0.11/thirdparty/argparse/tools/build.bat

  old	new
  1		call "C:\Program Files (x86)\Microsoft Visual Studio\2017\BuildTools\VC\Auxiliary\Build\vcvars64.bat"
  2
  3		cmake -B build -G Ninja -DCMAKE_BUILD_TYPE=Release -DARGPARSE_BUILD_TESTS=ON
  4		ninja -C build
  	1	call "C:\Program Files (x86)\Microsoft Visual Studio\2017\BuildTools\VC\Auxiliary\Build\vcvars64.bat"
  	2
  	3	cmake -B build -G Ninja -DCMAKE_BUILD_TYPE=Release -DARGPARSE_BUILD_TESTS=ON
  	4	ninja -C build

That is to say, a Windows build file's line endings changed from CRLF to LF, which is very frustrating because ​the file in their repository has not changed so this should absolutely not happen. Perhaps GitLab changed how they automatically generate tarballs. This is one of the perils of tying ports to automatically-generated tarballs and why we really don't want to do it and only do it when the developers do not give us any other option.

Updating to 7.0.11 was always intended to be a stopgap before updating to 8.x which will of course fix the issue since the new version will have different checksums. Until then we can ignore the inexplicable stealth update.