diff options
| author | Craig Jennings <c@cjennings.net> | 2026-05-26 14:47:01 -0500 |
|---|---|---|
| committer | Craig Jennings <c@cjennings.net> | 2026-05-26 14:47:01 -0500 |
| commit | 798b86faca84eaeba4f3f5ba1658b453603527de (patch) | |
| tree | 7d9dd661fb82c2bd00b4ab31e5fa889bcb8f8a09 /scripts | |
| parent | 5d3c5525f4b8bf742448d3a4be160f8c9ecdf4ec (diff) | |
| download | archsetup-798b86faca84eaeba4f3f5ba1658b453603527de.tar.gz archsetup-798b86faca84eaeba4f3f5ba1658b453603527de.zip | |
fix(archsetup): give git_install the same retry logic as pacman/aur
pacman_install and aur_install retry up to MAX_INSTALL_RETRIES. git_install only retried the clone once and never the build. I folded it into the same retry loop: it cleans the build dir between attempts and runs clone && make-install as one command list, so the captured exit code is the real failure rather than the if-statement's status (the trap retry_install documents).
One behavior change: a build that fails every attempt now warns and continues instead of being fatal on the re-clone path. That matches how pacman and aur failures are handled, so one source build that won't compile no longer aborts the whole install. It surfaces in the error summary instead.
Diffstat (limited to 'scripts')
0 files changed, 0 insertions, 0 deletions
