Re: [PATCH 2/6] builtin/update-index: end ODB transaction when --verbose is specified

From: Justin Tobler <jltobler@gmail.com>
Date: Sep 11, 2025, 15:34 UTC
Message-ID: <3cmqsfz5j2h36mw7yzwbrgfgchaettcnidvkiboqqktkm3dd5k@h6rmgjnqdmq2>
In-Reply-To: <aMJu7vPKfnxt9UiC@pks.im>

On 25/09/11 08:40AM, Patrick Steinhardt wrote:

Show 26 quoted lines

> On Tue, Sep 09, 2025 at 02:11:30PM -0500, Justin Tobler wrote:
> > With 23a3a303 (update-index: use the bulk-checkin infrastructure,
> > 2022-04-04), object database transactions were added to
> > git-update-index(1) to facilitate writing objects in bulk. With
> > transactions, newly added objects are instead written to a temporary
> > object directory and migrated to the primary object database upon
> > transaction commit.
> > 
> > When the --verbose option is specified, each individual object is
> > explicitly flushed via flush_odb_transaction() prior to reporting the
> > update. Flushing the object database transaction migrates pending
> > objects to the primary object database without marking the transaction
> > as complete. This is done so objects are immediately visible to
> > git-update-index(1) callers using the --verbose option and that rely on
> > parsing verbose output to know when objects are written.
> > 
> > As soon as verbose output is requested in git-update-index(1), all
> > subsequent object writes are flushed prior to being reported and thus no
> > longer benefit from being transactional. Furthermore, the mechanism to
> > flush a transaction without committing is rather awkward. Drop the call
> > to flush_odb_transaction() in favor of ending the transaction early when
> > the --verbose flag is encountered.
> 
> Okay, this interface feels somewhat weird indeed. If we now end the
> transaction early, does the transaction still serve any purpose at all?
> Like, do we use it to batch steps _before_ we start reporting stuff?

We only start reporting updates when the --verbose option is first encountered. Options are not all processed upfront. This means in the follow example:

  $ git update-index --add foo --add bar --verbose --stdin

both "foo" and "bar" are silently added via a transaction. After the --verbose option, subsequent updates are reported. At this point there is no reason for the transaction to continue as all subsequent object writes must be fully written before being reported. Thus the transaction is ended early.

> If the answer is "no", we might be able to just not create a transaction
> altogether.

We still use a transaction in the normal case when objects are written without --verbose enabled. This matches the existing behavior.

> If the answer is "yes" we should probably point out in the
> commit message that the transaction still has a purpose.

Yes, transactions still have a purpose, but only for object written before the --verbose option is first encountered. I'll try to clarify this in the commit message.

-Justin