Historical bug tracker for the Pacman package manager.
The pacman bug tracker has moved to gitlab:
https://gitlab.archlinux.org/pacman/pacman/-/issues
This tracker remains open for interaction with historical bugs during the transition period. Any new bugs reports will be closed without further action.
The pacman bug tracker has moved to gitlab:
https://gitlab.archlinux.org/pacman/pacman/-/issues
This tracker remains open for interaction with historical bugs during the transition period. Any new bugs reports will be closed without further action.
FS#53646 - pacman's output breaks bash, causes posix mode
Attached to Project:
Pacman
Opened by MechaLynx (MechaLynx) - Wednesday, 12 April 2017, 04:57 GMT
Last edited by Andrew Gregory (andrewgregory) - Wednesday, 12 April 2017, 13:09 GMT
Opened by MechaLynx (MechaLynx) - Wednesday, 12 April 2017, 04:57 GMT
Last edited by Andrew Gregory (andrewgregory) - Wednesday, 12 April 2017, 13:09 GMT
|
DetailsSummary and Info:
Allowing pacman to produce output can change bash's behavior, specifically breaking behavior when sourcing files. Redirecting output prevents this. Steps to Reproduce: - start a bash script that sets extglob - source a file that requires extglob to work - call pacman after sourcing the file - sourcing breaks Test code: https://gist.github.com/MechaLynx/12b6f8898c80004e0052c57bffc2c824 Output: http://imgur.com/zXP656h Tested on the live archiso distribution, booted from the latest iso (april 2017) |
This task depends upon
Closed by Andrew Gregory (andrewgregory)
Wednesday, 12 April 2017, 13:09 GMT
Reason for closing: Upstream
Wednesday, 12 April 2017, 13:09 GMT
Reason for closing: Upstream
edit - Some discussion on #bash on freenode has revealed that it is the environment exported to pacman due to `set -a' that's causing the issue, since pacman tries to import the environment twice and because `extglob' isn't exported, it fails the import. Making sure `set +a' is used before calling pacman should be a practical workaround. Thanks to twkm and geirha on #bash. Not sure if this counts as a pacman bug anymore, however, it does show that pacman spawns at least 2 new bash environments to do its printing, while it doesn't spawn any if printing is redirected. I guess it counts as a bug if this forking can be avoided, but I guess that's asking too much.