FS#47943 - busybox du does not support --apparent-size
Attached to Project:
Pacman
Opened by Alastair Hughes (pypi) - Friday, 29 January 2016, 19:16 GMT
Last edited by Allan McRae (Allan) - Saturday, 14 January 2017, 07:59 GMT
Opened by Alastair Hughes (pypi) - Friday, 29 January 2016, 19:16 GMT
Last edited by Allan McRae (Allan) - Saturday, 14 January 2017, 07:59 GMT
|
Details
busybox does not support GNU du's --apparent-size
argument.
This causes a failure to build a package (du fails, which causes an infinite loop). This has come up before: https://bugs.archlinux.org/task/10459 https://lists.archlinux.org/pipermail/pacman-dev/2010-November/011989.html Currently, the lack of support in other implementations means that this is worked around in the configure script. Relevant commits/code: https://gitorious.org/pacman-bsd/pacman-bsd/commit/cf3341fe591293a493bc925e0433a1f696b37d90?p=pacman-bsd:pacman-bsd.git;a=commitdiff;h=cf3341fe591293a493bc925e0433a1f696b37d90;hp=8b367d4441ba85b5548285d987afcfc84c4fcb3e https://projects.archlinux.org/pacman.git/tree/configure.ac#n335 Currently I can workaround this issue by removing the --apparent-size argument. |
This task depends upon
Closed by Allan McRae (Allan)
Saturday, 14 January 2017, 07:59 GMT
Reason for closing: Fixed
Additional comments about closing: Use DUPATH during configure. Git commit 36f456cc
Saturday, 14 January 2017, 07:59 GMT
Reason for closing: Fixed
Additional comments about closing: Use DUPATH during configure. Git commit 36f456cc
The attached file seems to work largely as expected, but du ignores hard links, while this counts them twice. It also uses ftw, which appears to be an extension; I suppose that this is less than ideal, at least from a portability perspective. I appreciate that it is most likely completely unusable, but I'd love some constructive comments.
Assuming I managed to write something usable, how would the result be integrated into pacman? I'm assuming that it would be a standalone tool installed to the libexec dir, but where would it sit in the source tree?
EDIT: Follow up (modified, but same general approach) patch posted: https://lists.archlinux.org/pipermail/pacman-dev/2016-August/021266.html