[PATCH] rev-list: honor --abbrev=<n> when doing --pretty=oneline

[PATCH] rev-list: honor --abbrev=<n> when doing --pretty=oneline

[Eric Wong]

This should make --pretty=oneline a whole lot more readable for
people using 80-column terminals.

Note that --abbrev=DEFAULT_ABBREV was on by default before, but
it only affected the printing of the Merge: header).  Let me
know if anybody doesn't want the default behavior to change.
Also note that --abbrev without arguments is not supported by
rev-list, but --no-abbrev is supported if you want the old
behavior.

Originally I made abbrev affect the commit sha1 output
regardless of the pretty setting, but that broke some tests and
I figured it's most/only useful for --pretty=oneline (at least
that's why *I* want it :)

Signed-off-by: Eric Wong <normalperson@yhbt.net>

---

 rev-list.c |    5 ++++-
 1 files changed, 4 insertions(+), 1 deletions(-)

c4da073e8256499950e25e2c20ea0b3ec4c29b46
diff --git a/rev-list.c b/rev-list.c
index 22141e2..392209d 100644
--- a/rev-list.c
+++ b/rev-list.c
@@ -52,7 +52,10 @@ static void show_commit(struct commit *c
 		fputs(commit_prefix, stdout);
 	if (commit->object.flags & BOUNDARY)
 		putchar('-');
-	fputs(sha1_to_hex(commit->object.sha1), stdout);
+	if (abbrev && commit_format == CMIT_FMT_ONELINE)
+		fputs(find_unique_abbrev(commit->object.sha1, abbrev), stdout);
+	else
+		fputs(sha1_to_hex(commit->object.sha1), stdout);
 	if (revs.parents) {
 		struct commit_list *parents = commit->parents;
 		while (parents) {
-- 
1.3.0.rc2.g454a-dirty

Re: [PATCH] rev-list: honor --abbrev=<n> when doing --pretty=oneline

[Junio C Hamano]

Eric Wong <normalperson@yhbt.net> writes:

> Note that --abbrev=DEFAULT_ABBREV was on by default before, but
> it only affected the printing of the Merge: header).  Let me
> know if anybody doesn't want the default behavior to change.

I've never felt need for abbreviating commit object names, so I
only had the abbrev variable to determine how the merge parents
are shown.  If you want to abbreviate the commit object names as
well, you _could_ do independent precision for parents and
commits, but that would be overkil.  So I'd rather see a switch
to turn abbreviation for commits on, perhaps like this:

        $ git-rev-list --pretty=oneline --abbrev-commit -n 3 master
        454a35b Add documentation for git-imap-send.
        ba3c937 blame.c: fix completely broken ancestry traversal.
        6cbd5d7 Tweaks to make asciidoc play nice.

        $ git-rev-list --pretty=oneline --abbrev=4 --abbrev-commit -n 3 master
        454a Add documentation for git-imap-send.
        ba3c9 blame.c: fix completely broken ancestry traversal.
        6cbd5 Tweaks to make asciidoc play nice.

Otherwise you might break Porcelains and people's scripts that
read from --pretty or --header output.

-- >8 --
diff --git a/rev-list.c b/rev-list.c
index 22141e2..1301502 100644
--- a/rev-list.c
+++ b/rev-list.c
@@ -30,6 +30,7 @@ static const char rev_list_usage[] =
 "    --unpacked\n"
 "    --header | --pretty\n"
 "    --abbrev=nr | --no-abbrev\n"
+"    --abbrev-commit\n"
 "  special purpose:\n"
 "    --bisect"
 ;
@@ -39,6 +40,7 @@ struct rev_info revs;
 static int bisect_list = 0;
 static int verbose_header = 0;
 static int abbrev = DEFAULT_ABBREV;
+static int abbrev_commit = 0;
 static int show_timestamp = 0;
 static int hdr_termination = 0;
 static const char *commit_prefix = "";
@@ -52,7 +54,10 @@ static void show_commit(struct commit *c
 		fputs(commit_prefix, stdout);
 	if (commit->object.flags & BOUNDARY)
 		putchar('-');
-	fputs(sha1_to_hex(commit->object.sha1), stdout);
+	if (abbrev_commit && abbrev)
+		fputs(find_unique_abbrev(commit->object.sha1, abbrev), stdout);
+	else
+		fputs(sha1_to_hex(commit->object.sha1), stdout);
 	if (revs.parents) {
 		struct commit_list *parents = commit->parents;
 		while (parents) {
@@ -317,6 +322,14 @@ int main(int argc, const char **argv)
 		}
 		if (!strcmp(arg, "--no-abbrev")) {
 			abbrev = 0;
+			continue;
+		}
+		if (!strcmp(arg, "--abbrev")) {
+			abbrev = DEFAULT_ABBREV;
+			continue;
+		}
+		if (!strcmp(arg, "--abbrev-commit")) {
+			abbrev_commit = 1;
 			continue;
 		}
 		if (!strncmp(arg, "--abbrev=", 9)) {

Re: [PATCH] rev-list: honor --abbrev=<n> when doing --pretty=oneline

[Eric Wong]

Junio C Hamano <junkio@cox.net> wrote:
> Eric Wong <normalperson@yhbt.net> writes:
> 
> > Note that --abbrev=DEFAULT_ABBREV was on by default before, but
> > it only affected the printing of the Merge: header).  Let me
> > know if anybody doesn't want the default behavior to change.
> 
> I've never felt need for abbreviating commit object names, so I
> only had the abbrev variable to determine how the merge parents
> are shown.  If you want to abbreviate the commit object names as
> well, you _could_ do independent precision for parents and
> commits, but that would be overkil.  So I'd rather see a switch
> to turn abbreviation for commits on, perhaps like this:
> 
>         $ git-rev-list --pretty=oneline --abbrev-commit -n 3 master
>         454a35b Add documentation for git-imap-send.
>         ba3c937 blame.c: fix completely broken ancestry traversal.
>         6cbd5d7 Tweaks to make asciidoc play nice.
> 
>         $ git-rev-list --pretty=oneline --abbrev=4 --abbrev-commit -n 3 master
>         454a Add documentation for git-imap-send.
>         ba3c9 blame.c: fix completely broken ancestry traversal.
>         6cbd5 Tweaks to make asciidoc play nice.
> 
> Otherwise you might break Porcelains and people's scripts that
> read from --pretty or --header output.
> 
> -- >8 --

Sounds good, I like your patch.  I'm not thrilled with the length of the
'--abbrev-commit' switch, but I guess that's what aliases are for :>

-- 
Eric Wong