Message ID | cover.1618835148.git.ps@pks.im (mailing list archive) |
---|---|
Headers | show |
Series | config: allow overriding global/system config | expand |
Patrick Steinhardt <ps@pks.im> writes: > this is the fifth version of my patch series to provide a way of > overriding the global system configuration. Hmm, this topic has been in next since April 15th. It is preferrable to fix things with incremental updates (the reasoning being that while in 'seen' the issues are "mistakes by the author alone", which may not necessarily benefit others to learn from, but what is in 'next' are supposed to have got enough review exposure, so any mistakes found are both of the author and reviewers, i.e. trickier than what is found before it hits 'next' and are more worth documenting with incremental "oops that was wrong for such and such reason, which even our reviews missed, and here is a fix"). But I'll have to go offline for a while soon, so let's make an exception, revert the merge of the previous round out of 'next', and queue this in 'seen'. Thanks.
On Mon, Apr 19, 2021 at 02:31:04PM +0200, Patrick Steinhardt wrote: > this is the fifth version of my patch series to provide a way of > overriding the global system configuration. > > Changes to v4: > > - Readded the call to `git_config_system()`, which I've previously > dropped by accident. I didn't move it into the new > `git_system_config()` function as it would change semantics of > `git config --system`. > > - Added a testcase which verifies that GIT_CONFIG_NOSYSTEM and > GIT_CONFIG_SYSTEM properly interact with each other: if > GIT_CONFIG_NOSYSTEM is set, no system-level configuration shall be > read. This is different than the tests for `git config --system` > which used to and still does ignore GIT_CONFIG_NOSYSTEM. > > - Small fixups for another testcase to drop needless redirects and > the `sane_unset` of GIT_CONFIG_NOSYSTEM. With the extra fixup to avoid unsetting GIT_CONFIG_NOSYSTEM for the whole test suite, this looks good to me. -Peff