From: Junio C Hamano <gitster@pobox.com>
To: Ronnie Sahlberg <sahlberg@google.com>
Cc: git@vger.kernel.org
Subject: Re: [PATCH v3 19/23] refs-be-files.c: add a backend method structure with transaction functions
Date: Tue, 26 Aug 2014 14:38:42 -0700 [thread overview]
Message-ID: <xmqq8umbgcd9.fsf@gitster.dls.corp.google.com> (raw)
In-Reply-To: <1408465847-30384-20-git-send-email-sahlberg@google.com> (Ronnie Sahlberg's message of "Tue, 19 Aug 2014 09:30:43 -0700")
Ronnie Sahlberg <sahlberg@google.com> writes:
> diff --git a/refs-be-files.c b/refs-be-files.c
> index e58a7e1..27eafd0 100644
> --- a/refs-be-files.c
> +++ b/refs-be-files.c
> ...
> +struct ref_be refs_files = {
> + files_transaction_begin,
> + files_transaction_update_sha1,
> + files_transaction_create_sha1,
> + files_transaction_delete_sha1,
> + files_transaction_update_reflog,
> + files_transaction_commit,
> + files_transaction_free,
> +};
> +
> +struct ref_be *refs = &refs_files;
> diff --git a/refs.c b/refs.c
> index 6b434ad..b8c942f 100644
> --- a/refs.c
> +++ b/refs.c
> ...
> +void transaction_free(struct ref_transaction *transaction)
> +{
> + return refs->transaction_free(transaction);
> +}
> diff --git a/refs.h b/refs.h
> index a14fc5d..4b669f5 100644
> --- a/refs.h
> +++ b/refs.h
> ...
> +struct ref_be {
> + transaction_begin_fn transaction_begin;
> + transaction_update_sha1_fn transaction_update_sha1;
> + transaction_create_sha1_fn transaction_create_sha1;
> + transaction_delete_sha1_fn transaction_delete_sha1;
> + transaction_update_reflog_fn transaction_update_reflog;
> + transaction_commit_fn transaction_commit;
> + transaction_free_fn transaction_free;
> +};
> +
> +extern struct ref_be *refs;
> +
> #endif /* REFS_H */
The overall structure is certainly nice, but this means you only can
LINK with one backend. Is that what we really want?
I would have expected something like this:
* In refs.c, there is a "static struct ref_be *the_refs_backend"
that points at the chosen singleton backend;
* Upon start-up, set_refs_backend() function that is exported from
refs.c can be used to set the_refs_backend;
* Each refs-be-frotz.c will export "struct ref_be refs_frotz" (or
perhaps "struct refs_be refs_be_frotz") to the outside world, so
that the start-up code can call set_refs_backend() with it.
* It is probably sensible to keep the_refs_backend default to
&refs_be_files.
next prev parent reply other threads:[~2014-08-26 21:38 UTC|newest]
Thread overview: 27+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-08-19 16:30 [PATCH v3 00/23] backend-struct-db Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 01/23] refs.c: create a public function for is_refname_available Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 03/23] refs.c: add a new refs.c file to hold all common refs code Ronnie Sahlberg
2014-08-26 21:31 ` Junio C Hamano
2014-08-26 22:00 ` Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 04/23] refs.c: move update_ref to refs.c Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 05/23] refs.c: move delete_ref to the common code Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 06/23] refs.c: move rename_ref " Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 07/23] refs.c: move read_ref_at to the common refs file Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 08/23] refs.c: move the hidden refs functions to the common code Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 09/23] refs.c: move dwim and friend functions to the common refs code Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 10/23] refs.c: move warn_if_dangling_symref* to the common code Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 11/23] refs.c: move read_ref, read_ref_full and ref_exists " Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 12/23] refs.c: move resolve_refdup to common Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 13/23] refs.c: move check_refname_component to the common code Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 14/23] refs.c: move is_branch " Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 15/23] refs.c: move names_conflict " Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 16/23] refs.c: move prettify_refname " Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 17/23] refs.c: move ref iterators " Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 18/23] refs.c: move head_ref_namespaced " Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 19/23] refs-be-files.c: add a backend method structure with transaction functions Ronnie Sahlberg
2014-08-26 21:38 ` Junio C Hamano [this message]
2014-08-26 22:34 ` Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 20/23] refs-be-files.c: add reflog backend methods Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 21/23] refs-be-files.c: add methods for misc ref operations Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 22/23] refs-be-files.c: add methods for head_ref* Ronnie Sahlberg
2014-08-19 16:30 ` [PATCH v3 23/23] refs-be-files.c: add methods for the ref iterators Ronnie Sahlberg
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=xmqq8umbgcd9.fsf@gitster.dls.corp.google.com \
--to=gitster@pobox.com \
--cc=git@vger.kernel.org \
--cc=sahlberg@google.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.