Thread (47 messages) 47 messages, 4 authors, 2021-12-11

Re: [PATCH 12/19] tests: fix broken &&-chains in `{...}` groups

From: Eric Sunshine <hidden>
Date: 2021-12-11 07:14:56

On Fri, Dec 10, 2021 at 4:29 AM Jeff King [off-list ref] wrote:
On Thu, Dec 09, 2021 at 12:11:08AM -0500, Eric Sunshine wrote:
quoted
Fix broken &&-chains in `{...}` groups in order to reduce the number of
possible lurking bugs.
Seems good. This is mostly stuff we don't expect to fail (mostly
"echo"), so I doubt they're important on their own. But getting a clean
state for the linter _is_ important.
In this patch, I think the only &&-chain which really matters (i.e.
could hide a genuine failure if broken) is t5515-fetch-merge-logic.sh.
Aside from most of these being unlikely to fail (`echo`), most of the
exit codes are being lost down pipes anyhow. As such, I had a hard
time justifying this patch since it exists mostly to satisfy the
linter which isn't smart enough to distinguish between the cases. What
ultimately convinced me that this patch was worthwhile was (1) that
there are legitimate cases, such as t5515-fetch-merge-logic.sh, where
we really do want to be told about the broken &&-chain, and (2) that
it's easier to have a single simple rule which we can point test
authors at ("chain all your test commands with `&&`") rather than
complex rules laying out cases when you do and don't need to maintain
the &&-chain.
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help