Re: [PATCH 4/6] revert: Make commit descriptions in insn sheet optional
- From
Junio C Hamano <gitster@pobox.com>
- Date
- Oct 10, 2011, 17:54 UTC
- Message-ID
- <7vwrccn34l.fsf@alter.siamese.dyndns.org>
- In-Reply-To
- <1318095407-26429-5-git-send-email-artagnon@gmail.com>
Ramkumar Ramachandra <artagnon@gmail.com> writes:
Show 52 quoted lines
> Change the instruction sheet format subtly so that a description of
> the commit after the object name is optional. As a result, an
> instruction sheet like this is now perfectly valid:
>
> pick 35b0426
> pick fbd5bbcbc2e
> pick 7362160f
>
> Suggested-by: Jonathan Nieder <jrnieder@gmail.com>
> Acked-by: Jonathan Nieder <jrnieder@gmail.com>
> Signed-off-by: Ramkumar Ramachandra <artagnon@gmail.com>
> ---
> builtin/revert.c | 19 ++++++++-----------
> t/t3510-cherry-pick-sequence.sh | 14 ++++++++++++++
> 2 files changed, 22 insertions(+), 11 deletions(-)
>
> diff --git a/builtin/revert.c b/builtin/revert.c
> index 6451089..aa6c34e 100644
> --- a/builtin/revert.c
> +++ b/builtin/revert.c
> @@ -692,26 +692,23 @@ static struct commit *parse_insn_line(char *start, struct replay_opts *opts)
> unsigned char commit_sha1[20];
> char sha1_abbrev[40];
> enum replay_action action;
> - int insn_len = 0;
> - char *p, *q;
> + const char *p, *q;
>
> + p = start;
> if (!prefixcmp(start, "pick ")) {
> action = CHERRY_PICK;
> - insn_len = strlen("pick");
> - p = start + insn_len + 1;
> + p += strlen("pick ");
> } else if (!prefixcmp(start, "revert ")) {
> action = REVERT;
> - insn_len = strlen("revert");
> - p = start + insn_len + 1;
> + p += strlen("revert ");
> } else
> return NULL;
>
> - q = strchr(p, ' ');
> - if (!q || q - p + 1 > sizeof(sha1_abbrev))
> + q = p + strcspn(p, " \n");
> + if (q - p + 1 > sizeof(sha1_abbrev))
> return NULL;
> - q++;
> -
> - strlcpy(sha1_abbrev, p, q - p);
> + memcpy(sha1_abbrev, p, q - p);
> + sha1_abbrev[q - p] = '\0';Since this is a part of clean-up series...
Do you even need to have a sha1_abbrev[] local array that is limited to 40 bytes here? The incoming _line_ is not "const char *start", so you should at least be able to temporarily terminate the commit object name with NUL (while remembering what byte there was before), give it to get_sha1(), and then restore the byte at the end before returning from this function.
A bonus point would be to introduce a variant of get_sha1() that can take (a pointer + len) not (a pointer to NUL terminated string). While I think that would be a right thing to do in the longer term, it is outside of the scope of this series.