From: Pranit Bauva <pranit.bauva@gmail.com>
To: Eric Sunshine <sunshine@sunshineco.com>
Cc: "Git List" <git@vger.kernel.org>,
"Christian Couder" <christian.couder@gmail.com>,
"Christian Couder" <chriscool@tuxfamily.org>,
"Lars Schneider" <larsxschneider@gmail.com>,
"Torsten Bögershausen" <tboegi@web.de>
Subject: Re: [PATCH v2 3/6] wrapper: move is_empty_file() from builtin/am.c
Date: Thu, 16 Jun 2016 00:10:03 +0530 [thread overview]
Message-ID: <CAFZEwPOkojqvivXpP7vq6WuMWPAjE1uXruoxsxH76LGT+1Eu4A@mail.gmail.com> (raw)
In-Reply-To: <CAPig+cTu1msxkjBZgLfy9g3+_VF5OzKNDbR0E0zVvopTNe2Q=g@mail.gmail.com>
Hey Eric,
On Wed, Jun 15, 2016 at 11:52 PM, Eric Sunshine <sunshine@sunshineco.com> wrote:
> On Wed, Jun 15, 2016 at 10:00 AM, Pranit Bauva <pranit.bauva@gmail.com> wrote:
>> is_empty_file() can help to refactor a lot of code. Also it is quite
>> helpful while converting shell scripts which use `test -s`. Since
>
> As justification, "can help to refactor a lot of code" is very
> nebulous. It would be better to give a concrete reason for moving the
> function, such as explaining that the functionality will be needed by
> the "git bisect" port to C.
Sure I can include that.
>> is_empty_file() is now a "library" function, its inappropriate to die() so
>> instead error_errno() is used to convey the message to stderr while the
>> appropriate boolean value is returned.
>>
>> Signed-off-by: Pranit Bauva <pranit.bauva@gmail.com>
>> ---
>> diff --git a/builtin/am.c b/builtin/am.c
>> @@ -30,22 +30,6 @@
>> /**
>> - * Returns 1 if the file is empty or does not exist, 0 otherwise.
>> - */
>> -static int is_empty_file(const char *filename)
>> -{
>> - struct stat st;
>> -
>> - if (stat(filename, &st) < 0) {
>> - if (errno == ENOENT)
>> - return 1;
>> - die_errno(_("could not stat %s"), filename);
>> - }
>> -
>> - return !st.st_size;
>> -}
>
> So, the original function die()'d for unexpected errors, but the
> rewrite does not. This is a big behavior change. To account for such a
> change in behavior I'd expect to see git-am updated to die() on its
> own for such failures, but no such changes are present in this patch.
> More about this below...
>
>> diff --git a/wrapper.c b/wrapper.c
>> @@ -696,3 +696,16 @@ void sleep_millisec(int millisec)
>> +int is_empty_file(const char *filename)
>> +{
>> + struct stat st;
>> +
>> + if (stat(filename, &st) < 0) {
>> + if (errno == ENOENT)
>> + return 1;
>> + error_errno(_("could not stat %s"), filename);
>
> Mental note: There is no 'return' in front of error_errno(), so the
> function does not exit here...
This is purposely as I want to keep this function to return only
boolean values ( 0 or 1).
>> + }
>> +
>> + return !st.st_size;
>> +}
>
> If stat() returns some error other than ENOENT, then the value of 'st'
> will be undefined, yet this return statement accesses its 'st_size'
> field, which is clearly a bad thing to do.
>
> You either need to return a designated value (such as -1) upon errors
> other than ENOENT (and update the documentation to mention -1) so that
> the caller can decided what to do, or die() as the original did. While
> it's true that die()'ing is not necessarily friendly in library code,
> it may be acceptable until such time that you find a caller which
> needs different behavior.
I didn't realize the complexity the patch carried with itself before.
I probably shouldn't fidget with am code right now, its a work better
left to who are converting the code to library code. I think its the
best fit for this situation to leave it as die()'ing.
Regards,
Pranit Bauva
next prev parent reply other threads:[~2016-06-15 18:40 UTC|newest]
Thread overview: 75+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-06-07 20:54 [PATCH 1/4] bisect--helper: `bisect_clean_state` shell function in C Pranit Bauva
2016-06-07 20:54 ` [PATCH 2/4] t6030: explicitly test for bisection cleanup Pranit Bauva
2016-06-07 23:21 ` Eric Sunshine
2016-06-08 8:07 ` Pranit Bauva
2016-06-08 8:17 ` Eric Sunshine
2016-06-08 10:20 ` Pranit Bauva
2016-06-07 20:54 ` [PATCH 3/4] dir: introduce file_size() to check the size of file Pranit Bauva
2016-06-08 7:37 ` Eric Sunshine
2016-06-08 7:57 ` Pranit Bauva
2016-06-08 8:13 ` Eric Sunshine
2016-06-08 10:03 ` Christian Couder
2016-06-08 8:17 ` Torsten Bögershausen
2016-06-08 13:08 ` Pranit Bauva
2016-06-12 10:44 ` Torsten Bögershausen
2016-06-13 6:21 ` Pranit Bauva
2016-06-07 20:54 ` [PATCH 4/4] bisect--helper: `bisect_reset` shell function in C Pranit Bauva
2016-06-08 7:59 ` Eric Sunshine
2016-06-08 9:51 ` Christian Couder
2016-06-08 9:53 ` Christian Couder
2016-06-08 17:50 ` Eric Sunshine
2016-06-08 13:20 ` Pranit Bauva
2016-06-08 17:53 ` Eric Sunshine
2016-06-08 18:04 ` Pranit Bauva
2016-06-07 22:31 ` [PATCH 1/4] bisect--helper: `bisect_clean_state` " Eric Sunshine
2016-06-08 1:51 ` Eric Sunshine
2016-06-08 7:46 ` Pranit Bauva
2016-06-08 8:02 ` Eric Sunshine
2016-06-08 8:09 ` Pranit Bauva
2016-06-08 9:41 ` Christian Couder
2016-06-08 17:59 ` Eric Sunshine
2016-06-08 18:04 ` Pranit Bauva
2016-06-15 14:00 ` [PATCH v2 0/6] convert various shell functions in git-bisect to C Pranit Bauva
2016-06-15 14:00 ` [PATCH v2 1/6] bisect--helper: `bisect_clean_state` shell function in C Pranit Bauva
2016-06-15 18:04 ` Eric Sunshine
2016-06-15 18:47 ` Pranit Bauva
2016-06-15 20:22 ` Eric Sunshine
2016-06-15 14:00 ` [PATCH v2 2/6] t6030: explicitly test for bisection cleanup Pranit Bauva
2016-06-15 14:00 ` [PATCH v2 3/6] wrapper: move is_empty_file() from builtin/am.c Pranit Bauva
2016-06-15 18:12 ` Junio C Hamano
2016-06-15 18:15 ` Pranit Bauva
2016-06-15 18:22 ` Eric Sunshine
2016-06-15 18:40 ` Pranit Bauva [this message]
2016-06-15 14:00 ` [PATCH v2 4/6] bisect--helper: `bisect_reset` shell function in C Pranit Bauva
2016-06-15 21:05 ` Eric Sunshine
2016-06-16 19:06 ` Pranit Bauva
2016-06-15 14:00 ` [PATCH v2 5/6] bisect--helper: `is_expected_rev` & `check_expected_revs` " Pranit Bauva
2016-06-15 21:14 ` Eric Sunshine
2016-06-16 19:05 ` Pranit Bauva
2016-06-16 19:16 ` Eric Sunshine
2016-06-16 19:25 ` Pranit Bauva
2016-06-16 20:47 ` Christian Couder
2016-06-17 12:49 ` Pranit Bauva
2016-06-15 14:00 ` [PATCH v2 6/6] bisect--helper: `bisect_write` " Pranit Bauva
2016-06-16 18:55 ` Eric Sunshine
2016-06-16 19:01 ` Pranit Bauva
2016-06-16 20:38 ` Christian Couder
2016-06-17 13:10 ` Pranit Bauva
2016-06-15 17:53 ` [PATCH v2 0/6] convert various shell functions in git-bisect to C Eric Sunshine
2016-06-15 18:09 ` Pranit Bauva
2016-06-26 12:23 ` [PATCH v3 " Pranit Bauva
2016-07-06 20:25 ` [PATCH v4 " Pranit Bauva
2016-07-06 20:25 ` [PATCH v4 1/6] bisect--helper: `bisect_clean_state` shell function in C Pranit Bauva
2016-07-06 20:25 ` [PATCH v4 2/6] t6030: explicitly test for bisection cleanup Pranit Bauva
2016-07-11 19:16 ` Junio C Hamano
2016-07-06 20:25 ` [PATCH v4 3/6] wrapper: move is_empty_file() and rename it as is_empty_or_missing_file() Pranit Bauva
2016-07-06 20:25 ` [PATCH v4 4/6] bisect--helper: `bisect_reset` shell function in C Pranit Bauva
2016-07-06 20:25 ` [PATCH v4 5/6] bisect--helper: `is_expected_rev` & `check_expected_revs` " Pranit Bauva
2016-07-06 20:25 ` [PATCH v4 6/6] bisect--helper: `bisect_write` " Pranit Bauva
2016-07-11 19:19 ` [PATCH v4 0/6] convert various shell functions in git-bisect to C Junio C Hamano
2016-06-26 12:23 ` [PATCH v3 1/6] bisect--helper: `bisect_clean_state` shell function in C Pranit Bauva
2016-06-26 12:23 ` [PATCH v3 2/6] t6030: explicitly test for bisection cleanup Pranit Bauva
2016-06-26 12:23 ` [PATCH v3 3/6] wrapper: move is_empty_file() and rename it as is_empty_or_missing_file() Pranit Bauva
2016-06-26 12:23 ` [PATCH v3 4/6] bisect--helper: `bisect_reset` shell function in C Pranit Bauva
2016-06-26 12:23 ` [PATCH v3 5/6] bisect--helper: `is_expected_rev` & `check_expected_revs` " Pranit Bauva
2016-06-26 12:23 ` [PATCH v3 6/6] bisect--helper: `bisect_write` " Pranit Bauva
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
List information: http://vger.kernel.org/majordomo-info.html
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CAFZEwPOkojqvivXpP7vq6WuMWPAjE1uXruoxsxH76LGT+1Eu4A@mail.gmail.com \
--to=pranit.bauva@gmail.com \
--cc=chriscool@tuxfamily.org \
--cc=christian.couder@gmail.com \
--cc=git@vger.kernel.org \
--cc=larsxschneider@gmail.com \
--cc=sunshine@sunshineco.com \
--cc=tboegi@web.de \
/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.
Code repositories for project(s) associated with this public inbox
https://80x24.org/mirrors/git.git
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).