From: Junio C Hamano <gitster@pobox.com>
To: Christian Couder <chriscool@tuxfamily.org>
Cc: git@vger.kernel.org, Johan Herland <johan@herland.net>,
Josh Triplett <josh@joshtriplett.org>,
Thomas Rast <tr@thomasrast.ch>,
Michael Haggerty <mhagger@alum.mit.edu>,
Dan Carpenter <dan.carpenter@oracle.com>,
Greg Kroah-Hartman <greg@kroah.com>, Jeff King <peff@peff.net>,
Eric Sunshine <sunshine@sunshineco.com>,
Ramsay Jones <ramsay@ramsay1.demon.co.uk>
Subject: Re: [PATCH v7 03/11] trailer: read and process config information
Date: Fri, 07 Mar 2014 13:24:32 -0800 [thread overview]
Message-ID: <xmqqmwh1u2gf.fsf@gitster.dls.corp.google.com> (raw)
In-Reply-To: <20140306221409.29648.10706.chriscool@tuxfamily.org> (Christian Couder's message of "Thu, 06 Mar 2014 23:14:00 +0100")
Christian Couder <chriscool@tuxfamily.org> writes:
> This patch implements reading the configuration
> to get trailer information, and then processing
> it and storing it in a doubly linked list.
"Read and process the ...", perhaps?
> The config information is stored in the list
> whose first item is pointed to by:
>
> static struct trailer_item *first_conf_item;
If feels somewhat strange if a doubly linked list has only the first
pointer without the last pointer, unless the previous field of the
first one by convention points at the last one, forming a cycle
(which I think is a reasonable thing to do, as a desire for a quick
access to the top and the bottom is often a reason to use doubly
linked list---I didn't check if you implement it that way, though).
> +static int set_where(struct conf_info *item, const char *value)
> +{
> + if (!strcasecmp("after", value))
> + item->where = WHERE_AFTER;
> + else if (!strcasecmp("before", value))
> + item->where = WHERE_BEFORE;
> + else
> + return 1;
Please follow the usual convention of returning a negative value
upon error, unless there is a compelling reason not to do so.
Do we really want to do "strcasecmp" here? Are some values case
sensitive and we only allow "after", "AfTer", "AFTER", etc. as case
insensitive keywords? If all values are expected to be case
insensitive, wouldn't it be better to downcase in the caller (just
like we do in the config parser) and use strcmp() against lower case
constants?
All of the comments applies equally to other functions.
> +enum trailer_info_type { TRAILER_KEY, TRAILER_COMMAND, TRAILER_WHERE,
> + TRAILER_IF_EXISTS, TRAILER_IF_MISSING };
> +
> +static int set_name_and_type(const char *conf_key, const char *suffix,
> + enum trailer_info_type type,
> + char **pname, enum trailer_info_type *ptype)
> +{
> + int ret = ends_with(conf_key, suffix);
> + if (ret) {
> + *pname = xstrndup(conf_key, strlen(conf_key) - strlen(suffix));
> + *ptype = type;
> + }
> + return ret;
> +}
This implementation, together with the caller that makes many calls
to it, looks overly inefficient (both runtime- and reviewtime-wise),
doesn't it?
How about having the caller upfront find the last dot before
checking the name-and-type to avoid calling ends_with() so many
times unnecessarily?
Also, wouldn't it be better to make the caller more table-driven,
i.e.
static struct {
const char *name;
enum trailer_kinfo_type type;
} trailer_config_items[] = {
{ "key", TRAILER_KEY },
...
};
in the caller's scope, and then
const char *trailer_item, *variable_name;
trailer_item = skip_prefix(conf_key, "trailer.");
if (!trailer_item)
return 0;
variable_name = strrchr(trailer_item, '.');
if (!variable_name) {
... trailer_item is a two-level variable name.
... Handle it in whatever way.
return 0;
}
variable_name++;
for (i = 0; i < ARRAY_SIZE(trailer_config_items); i++) {
if (strcmp(trailer_config_items[i].name, variable_name))
continue;
name = xstrdup(...);
type = trailer_config_items[i].type;
goto found;
}
/* Unknown trailer.<item>.variable_name */
return 0;
found:
... do whatever depending on the type ...
or something?
> +static struct trailer_item *get_conf_item(const char *name)
> +{
> + struct trailer_item *item;
> + struct trailer_item *previous;
> +
> + /* Look up item with same name */
> + for (previous = NULL, item = first_conf_item;
> + item;
> + previous = item, item = item->next) {
> + if (!strcasecmp(item->conf.name, name))
> + return item;
> + }
> +
> + /* Item does not already exists, create it */
> + item = xcalloc(sizeof(struct trailer_item), 1);
> + item->conf.name = xstrdup(name);
> +
> + if (!previous)
> + first_conf_item = item;
> + else {
> + previous->next = item;
> + item->previous = previous;
> + }
> +
> + return item;
> +}
> +
> +static int git_trailer_config(const char *conf_key, const char *value, void *cb)
> +{
> + if (starts_with(conf_key, "trailer.")) {
> + const char *orig_conf_key = conf_key;
> + struct trailer_item *item;
> + struct conf_info *conf;
> + char *name;
> + enum trailer_info_type type;
> +
> + conf_key += 8;
> + if (!set_name_and_type(conf_key, ".key", TRAILER_KEY, &name, &type) &&
> + !set_name_and_type(conf_key, ".command", TRAILER_COMMAND, &name, &type) &&
> + !set_name_and_type(conf_key, ".where", TRAILER_WHERE, &name, &type) &&
> + !set_name_and_type(conf_key, ".ifexists", TRAILER_IF_EXISTS, &name, &type) &&
> + !set_name_and_type(conf_key, ".ifmissing", TRAILER_IF_MISSING, &name, &type))
> + return 0;
> +
> + item = get_conf_item(name);
> + conf = &item->conf;
> + free(name);
> +
> + switch (type) {
> + case TRAILER_KEY:
> + if (conf->key)
> + warning(_("more than one %s"), orig_conf_key);
> + conf->key = xstrdup(value);
> + break;
> + case TRAILER_COMMAND:
> + if (conf->command)
> + warning(_("more than one %s"), orig_conf_key);
> + conf->command = xstrdup(value);
> + break;
> + case TRAILER_WHERE:
> + if (set_where(conf, value))
> + warning(_("unknown value '%s' for key '%s'"), value, orig_conf_key);
> + break;
> + case TRAILER_IF_EXISTS:
> + if (set_if_exists(conf, value))
> + warning(_("unknown value '%s' for key '%s'"), value, orig_conf_key);
> + break;
> + case TRAILER_IF_MISSING:
> + if (set_if_missing(conf, value))
> + warning(_("unknown value '%s' for key '%s'"), value, orig_conf_key);
> + break;
> + default:
> + die("internal bug in trailer.c");
> + }
> + }
> + return 0;
> +}
next prev parent reply other threads:[~2014-03-07 21:24 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-03-06 22:14 [PATCH v7 03/11] trailer: read and process config information Christian Couder
2014-03-07 21:24 ` Junio C Hamano [this message]
2014-03-07 21:59 ` David Kastrup
2014-03-07 22:16 ` Junio C Hamano
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=xmqqmwh1u2gf.fsf@gitster.dls.corp.google.com \
--to=gitster@pobox.com \
--cc=chriscool@tuxfamily.org \
--cc=dan.carpenter@oracle.com \
--cc=git@vger.kernel.org \
--cc=greg@kroah.com \
--cc=johan@herland.net \
--cc=josh@joshtriplett.org \
--cc=mhagger@alum.mit.edu \
--cc=peff@peff.net \
--cc=ramsay@ramsay1.demon.co.uk \
--cc=sunshine@sunshineco.com \
--cc=tr@thomasrast.ch \
/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.