pkg_rpm produces RPMs with clamped timestamps even when not requested
#486
Comments
nacl
pushed a commit
to nacl/rules_pkg
that referenced
this issue
Dec 14, 2021
…ularity cleanup The logic for `pkg_rpm`'s `source_date_epoch` attribute should apply only when set. This change updates the logic so that a negative value for `source_date_epoch` results in no `--source_date_epoch` being passed to `make_rpm.py`. The default for it is now a negative value, enforcing this feature. A regression test was also provided. While writing it, I noticed a few quirks in `rpm` that impacted test execution. These were also addressed, although not in the best way. For more context, see bazelbuild#486. Fixes bazelbuild#486
nacl
pushed a commit
to nacl/rules_pkg
that referenced
this issue
Dec 14, 2021
…ty cleanup The logic for `pkg_rpm`'s `source_date_epoch` attribute should apply only when set. This change updates the logic so that a negative value for `source_date_epoch` results in no `--source_date_epoch` being passed to `make_rpm.py`. `None` does not seem to be a valid option. The default is now a negative number, thus disabling this feature unless explicitly requested. A regression test was also provided. While writing it, I noticed a few quirks in `rpm` that impacted test execution. These were also addressed, although perhaps not in the best way. For more context, see bazelbuild#486. Fixes bazelbuild#486
This was referenced Dec 14, 2021
Closed
nacl
pushed a commit
that referenced
this issue
Dec 15, 2021
…ty cleanup (#487) The logic for `pkg_rpm`'s `source_date_epoch` attribute should apply only when set. This change updates the logic so that a negative value for `source_date_epoch` results in no `--source_date_epoch` being passed to `make_rpm.py`. `None` does not seem to be a valid option. The default is now a negative number, thus disabling this feature unless explicitly requested. A regression test was also provided. While writing it, I noticed a few quirks in `rpm` that impacted test execution. These were also addressed, although perhaps not in the best way. For more context, see #486. NOTE: this does not address the similar issue impacting the legacy `pkg_rpm` rule. This is discussed in #488. Fixes #486.
nacl
added
P3
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
We've been doing work to port our old copies of
pkg_rpmto use the rules from upstream, an internal user noticed that the RPMs were installing files with "0" mtimes.This can wreak havoc with some update processes. Root cause looks to be the conditional that configures SOURCE_DATE_EPOCH behavior in
make_rpm.pyseeing a string"0"(which is truthy) instead of integer0(which is falsey). This was introduced in 06a2a23.The text was updated successfully, but these errors were encountered: