Re: What's cooking in git.git (Mar 2020, #10; Tue, 31)

From: Taylor Blau <me@ttaylorr.com>
Date: Apr 3, 2020, 20:04 UTC
Message-ID: <20200403200418.GA58491@syl.local>
In-Reply-To: <xmqq4ku46s8z.fsf@gitster.c.googlers.com>

Hi Junio,

On Tue, Mar 31, 2020 at 04:04:28PM -0700, Junio C Hamano wrote:

Show 11 quoted lines

> * tb/commit-graph-split-merge (2020-03-24) 3 commits
>   (merged to 'next' on 2020-03-31 at 2183baf09c)
>  + builtin/commit-graph.c: support '--input=graphed'
>  + builtin/commit-graph.c: introduce '--input=<source>'
>  + builtin/commit-graph.c: support '--split[=<strategy>]'
>
>  The code to write out the commit-graph has been taught a few
>  options to control if the resulting graph chains should be merged
>  or a single new incremental graph is created.
>
>  Will merge to 'master'.

I know that this series has been cooking for a while (and that I've nagged you about merging it down to master), but I think I may ask you to temporarily eject it until I can send some more patches on top.

In particular, I think that it may be worthwhile to get rid of '--split=merge-all' and '--input=graphed' in favor of some new patches that we've been using to introduce a '--split=replace' mode, which discards the existing graph completely, but writes out a new length-1 incremental commit graph.

I'm not sure if there is general interest in a '--split=merge-all' or '--input=graphed', and if there is I'd be happy to keep those patches around, but we've found them to be less useful than the current options that we're using.

I'll send the '--no-check-oids' patches separately, which I don't think needs to hold this up.

Thanks, Taylor