Re: [PATCH] wt-status: expand, not dwim, a "detached from" ref

From: Junio C Hamano <gitster@pobox.com>
Date: Aug 27, 2020, 02:10 UTC
Message-ID: <xmqqtuwosusc.fsf@gitster.c.googlers.com>
In-Reply-To: <20200827014723.GA750502@google.com>

Jonathan Nieder <jrnieder@gmail.com> writes:

Show 18 quoted lines

> Given the context above, two possibilities seem appealing:
>
>  A. Like you're hinting, could dwim_ref get a variant that returns -1
>     instead of die()ing on failure?  That way, we could fulfill the
>     intent described in b397ea48:
>
> 	When it cannot figure out the original ref, it shows an abbreviated
> 	SHA-1.
>
>  B. Alternatively, in the @{u} case could we switch together the <old>
>     and <new> pieces of the context from the
>
> 	checkout: moving from master to @{upstream}
>
>     reflog line to make "master@{upstream}"?  It's still possible for
>     the upstream to have changed since then, but at least in the
>     common case this would match the lookup that happened at checkout
>     time.

Ah, blast from the past.  Thanks for resurrecting.

If we are allowed to change what goes to reflog, can we do even better than recording master@{upstream} at the time "checkout" records the HEAD movement? "checkout: moving from next to master" would be far better than "moving from next to next@{upstream}" or "moving from next to @{upstream}".

Can we even change the phrasing altogether, like "checkout: moving from next to detached e9b77c..."? That would produce even more precise report.