blame: (dim rep. metadata lines or fields, decay date coloring)

RFC PATCH Superseded Stefan Beller 3 participants 17 messages Nov 10, 2017 – Feb 8, 2018 Documentation

Share & tools

Stefan Beller

author

2017-11-10 01:10 UTC #0 lore raw

cover blame: (dim rep. metadata lines or fields, decay date coloring)

This is picking up [1], but presenting it in another approach, as I realized these are orthogonal features:
* dimming repeated lines/fields of information
* giving a quick visual information how old (as a proxy for 'well tested') a line of code is.

Both features are configurable.

Any feedback welcome.

Thanks,
Stefan

[1] https://public-inbox.org/git/[email protected]/

Stefan Beller (4):
color.h: modernize header
builtin/blame: dim uninteresting metadata builtin/blame: add option to color metadata fields separately builtin/blame: highlight recently changed lines

 Documentation/config.txt |  25 ++++++
 builtin/blame.c          | 216 ++++++++++++++++++++++++++++++++++++++++++-----
 color.c                  |   2 -
 color.h                  |  49 ++++++++---
 t/t8012-blame-colors.sh  |  56 ++++++++++++
 5 files changed, 313 insertions(+), 35 deletions(-)
 create mode 100755 t/t8012-blame-colors.sh

Signature · 1 line

2.15.0.128.gcadd42da22

Stefan Beller

author

2017-11-10 01:10 UTC #1 lore raw

PATCH 1/4 color.h: modernize header

Signed-off-by Stefan Beller <[email protected]>

---
 color.c |  2 --
 color.h | 48 +++++++++++++++++++++++++++++++++++-------------
 2 files changed, 35 insertions(+), 15 deletions(-)

color.c +0 -2

index 9a9261ac16..e43da3ce53 100644
@@ -400,8 +400,6 @@ static int color_vfprintf(FILE *fp, const char *color, const char *fmt,
 	return r;
 }
 
-
-
 int color_fprintf(FILE *fp, const char *color, const char *fmt, ...)
 {
 	va_list args;

color.h +35 ~~-13~~

61 lines not on this page · raw message

Signature · 1 line

2.15.0.128.gcadd42da22

Stefan Beller

author

2017-11-10 01:10 UTC #2 lore raw

PATCH 2/4 builtin/blame: dim uninteresting metadata

In reply to Stefan Beller

When using git-blame lots of lines contain redundant information, for example in hunks that consist of multiple lines, the metadata (commit name, author, date) are repeated. A reader may not be interested in those, so offer an option to color the information that is repeated from the previous line differently.

Signed-off-by Stefan Beller <[email protected]>

---
 Documentation/config.txt |  5 ++++
 builtin/blame.c          | 68 ++++++++++++++++++++++++++++++++++++++----------
 color.h                  |  1 +
 t/t8012-blame-colors.sh  | 18 +++++++++++++
 4 files changed, 78 insertions(+), 14 deletions(-)
 create mode 100755 t/t8012-blame-colors.sh

Documentation/config.txt +5 -0

index 5f0d62753d..85dfdc6a9b 100644
@@ -1189,6 +1189,11 @@ color.status.<slot>::
 	status short-format), or
 	`unmerged` (files which have unmerged changes).
 
+color.blame.repeatedMeta::
+	Use the customized color for the part of git-blame output that
+	is repeated meta information per line (such as commit id,
+	author name, date and timezone). Defaults to dark gray.
+
 color.ui::
 	This variable determines the default value for variables such
 	as `color.diff` and `color.grep` that control the use of color

builtin/blame.c +54 ~~-14~~

142 lines not on this page · raw message

color.h +1 -0

index 6cd632c0d8..fce5588df9 100644
@@ -30,6 +30,7 @@ struct strbuf;
 #define GIT_COLOR_BLUE		"\033[34m"
 #define GIT_COLOR_MAGENTA	"\033[35m"
 #define GIT_COLOR_CYAN		"\033[36m"
+#define GIT_COLOR_DARK		"\033[1;30m"
 #define GIT_COLOR_BOLD_RED	"\033[1;31m"
 #define GIT_COLOR_BOLD_GREEN	"\033[1;32m"
 #define GIT_COLOR_BOLD_YELLOW	"\033[1;33m"

t/t8012-blame-colors.sh new file +18 -0

new file mode 100755
index 0000000000..b2de03b777
@@ -0,0 +1,18 @@
+#!/bin/sh
+
+test_description='colored git blame'
+. ./test-lib.sh
+
+PROG='git blame -c'
+. "$TEST_DIRECTORY"/annotate-tests.sh
+
+test_expect_success 'colored blame colors continuous lines' '
+	git blame --abbrev=12 --color-lines hello.c >actual.raw &&
+	test_decode_color <actual.raw >actual &&
+	grep "<BOLD;BLACK>(F" actual > F.expect &&
+	grep "<BOLD;BLACK>(H" actual > H.expect &&
+	test_line_count = 2 F.expect &&
+	test_line_count = 3 H.expect
+'
+
+test_done

Signature · 1 line

2.15.0.128.gcadd42da22

Stefan Beller

author

2017-11-10 01:10 UTC #3 lore raw

PATCH 3/4 builtin/blame: add option to color metadata fields separately

In reply to Stefan Beller

Unlike the previous commit, this dims colors for each metadata field individually.

Signed-off-by Stefan Beller <[email protected]>

---
 builtin/blame.c         | 82 +++++++++++++++++++++++++++++++++++++++++++------
 t/t8012-blame-colors.sh | 38 +++++++++++++++++++++++
 2 files changed, 111 insertions(+), 9 deletions(-)

builtin/blame.c +73 -9

index d15e901357..9d460597a2 100644
@@ -323,6 +323,7 @@ static const char *format_time(timestamp_t time, const char *tz_str,
 #define OUTPUT_SHOW_EMAIL	0400
 #define OUTPUT_LINE_PORCELAIN 	01000
 #define OUTPUT_COLOR_LINE	02000
+#define OUTPUT_COLOR_FIELDS	04000
 
 static void emit_porcelain_details(struct blame_origin *suspect, int repeat)
 {

@@ -370,6 +371,33 @@ static void emit_porcelain(struct blame_scoreboard *sb, struct blame_entry *ent,
 		putchar('\n');
 }
 
+static int had_same_field_previously(int opt, int field,
+			  struct blame_entry *ent,
+			  struct blame_entry *prev)
+{
+	struct commit_info ci, prev_ci;
+
+	switch (field) {
+	case OUTPUT_SHOW_SCORE:
+		return ent->score == prev->score;
+	case OUTPUT_SHOW_NAME:
+		return prev->suspect &&
+			!strcmp(ent->suspect->path, prev->suspect->path);
+	case OUTPUT_SHOW_NUMBER:
+		return ent->s_lno == prev->s_lno + prev->num_lines - 1;
+
+	case OUTPUT_NO_AUTHOR:
+		get_commit_info(ent->suspect->commit, &ci, 1);
+		get_commit_info(prev->suspect->commit, &prev_ci, 1);
+		return ((opt & OUTPUT_SHOW_EMAIL) &&
+			!strcmp(ci.author_mail.buf, prev_ci.author_mail.buf)) ||
+			!strcmp(ci.author.buf, prev_ci.author.buf);
+	default:
+		BUG("unknown field");
+	}
+	return 0;
+}
+
 static inline void colors_unset(const char **use_color, const char **reset_color)
 {
 	*use_color = "";

@@ -382,17 +410,36 @@ static inline void colors_set(const char **use_color, const char **reset_color)
 	*reset_color = GIT_COLOR_RESET;
 }
 
+static void setup_field_color(int opt, int cnt, int field,
+			      struct blame_entry *ent,
+			      struct blame_entry *prev,
+			      const char **use_color,
+			      const char **reset_color)
+{
+	if (!(opt & OUTPUT_COLOR_FIELDS))
+		return;
+
+	if ((cnt > 0 ||
+	     (prev && had_same_field_previously(opt, field, ent, prev))))
+		colors_set(use_color, reset_color);
+	else
+		colors_unset(use_color, reset_color);
+}
+
 static void setup_line_color(int opt, int cnt,
 			     const char **use_color,
 			     const char **reset_color)
 {
 	colors_unset(use_color, reset_color);
 
-	if ((opt & OUTPUT_COLOR_LINE) && cnt > 0)
+	if ((opt & (OUTPUT_COLOR_LINE | OUTPUT_COLOR_FIELDS)) && cnt > 0)
 		colors_set(use_color, reset_color);
 }
 
-static void emit_other(struct blame_scoreboard *sb, struct blame_entry *ent, int opt)
+static void emit_other(struct blame_scoreboard *sb,
+		       struct blame_entry *ent,
+		       struct blame_entry *prev,
+		       int opt)
 {
 	int cnt;
 	const char *cp;

@@ -432,18 +479,27 @@ static void emit_other(struct blame_scoreboard *sb, struct blame_entry *ent, int
 					   show_raw_time),
 			       ent->lno + 1 + cnt);
 		} else {
-			if (opt & OUTPUT_SHOW_SCORE)
+			if (opt & OUTPUT_SHOW_SCORE) {
+				setup_field_color(opt, cnt, OUTPUT_SHOW_SCORE,
+						  ent, prev, &col, &rcol);
 				printf(" %s%*d %02d%s", col,
 				       max_score_digits, ent->score,
 				       ent->suspect->refcnt, rcol);
-			if (opt & OUTPUT_SHOW_NAME)
+			}
+			if (opt & OUTPUT_SHOW_NAME) {
+				setup_field_color(opt, cnt, OUTPUT_SHOW_NAME,
+						   ent, prev, &col, &rcol);
 				printf(" %s%-*.*s%s", col, longest_file,
 						      longest_file,
 						      suspect->path,
 						      rcol);
-			if (opt & OUTPUT_SHOW_NUMBER)
+			}
+			if (opt & OUTPUT_SHOW_NUMBER) {
+				setup_field_color(opt, cnt, OUTPUT_SHOW_NUMBER,
+						  ent, prev, &col, &rcol);
 				printf(" %s%*d%s", col, max_orig_digits,
 				       ent->s_lno + 1 + cnt, rcol);
+			}
 			if (!(opt & OUTPUT_NO_AUTHOR)) {
 				const char *name;
 				int pad;

@@ -452,6 +508,8 @@ static void emit_other(struct blame_scoreboard *sb, struct blame_entry *ent, int
 				else
 					name = ci.author.buf;
 				pad = longest_author - utf8_strwidth(name);
+				setup_field_color(opt, cnt, OUTPUT_NO_AUTHOR,
+						  ent, prev, &col, &rcol);
 				printf(" %s(%s%*s %10s%s", col,
 				       name, pad, "",
 				       format_time(ci.author_time,

@@ -477,7 +535,7 @@ static void emit_other(struct blame_scoreboard *sb, struct blame_entry *ent, int
 
 static void output(struct blame_scoreboard *sb, int option)
 {
-	struct blame_entry *ent;
+	struct blame_entry *ent, *prev = NULL;
 
 	if (option & OUTPUT_PORCELAIN) {
 		for (ent = sb->ent; ent; ent = ent->next) {

@@ -499,7 +557,8 @@ static void output(struct blame_scoreboard *sb, int option)
 		if (option & OUTPUT_PORCELAIN)
 			emit_porcelain(sb, ent, option);
 		else {
-			emit_other(sb, ent, option);
+			emit_other(sb, ent, prev, option);
+			prev = ent;
 		}
 	}
 }

@@ -717,6 +776,7 @@ int cmd_blame(int argc, const char **argv, const char *prefix)
 		OPT_BIT('e', "show-email", &output_option, N_("Show author email instead of name (Default: off)"), OUTPUT_SHOW_EMAIL),
 		OPT_BIT('w', NULL, &xdl_opts, N_("Ignore whitespace differences"), XDF_IGNORE_WHITESPACE),
 		OPT_BIT(0, "color-lines", &output_option, N_("color redundant metadata from previous line"), OUTPUT_COLOR_LINE),
+		OPT_BIT(0, "color-fields", &output_option, N_("color redundant metadata fields from previous line"), OUTPUT_COLOR_FIELDS),
 
 		/*
 		 * The following two options are parsed by parse_revision_opt()

@@ -775,6 +835,10 @@ int cmd_blame(int argc, const char **argv, const char *prefix)
 	DIFF_OPT_CLR(&revs.diffopt, FOLLOW_RENAMES);
 	argc = parse_options_end(&ctx);
 
+	if ((output_option & OUTPUT_COLOR_LINE) &&
+	    (output_option & OUTPUT_COLOR_FIELDS))
+		die(_("cannot ask for colored lines and fields at the same time"));
+
 	if (incremental || (output_option & OUTPUT_PORCELAIN)) {
 		if (show_progress > 0)
 			die(_("--progress can't be used with --incremental or porcelain formats"));

@@ -978,8 +1042,8 @@ int cmd_blame(int argc, const char **argv, const char *prefix)
 
 	if (!(output_option & OUTPUT_PORCELAIN)) {
 		find_alignment(&sb, &output_option);
-		if ((output_option & OUTPUT_COLOR_LINE) &&
-		    !repeated_meta_color)
+		if (!repeated_meta_color &&
+		    (output_option & (OUTPUT_COLOR_LINE | OUTPUT_COLOR_FIELDS)))
 			repeated_meta_color = GIT_COLOR_DARK;
 	}

t/t8012-blame-colors.sh +38 -0

index b2de03b777..be491bde9a 100755
@@ -15,4 +15,42 @@ test_expect_success 'colored blame colors continuous lines' '
 	test_line_count = 3 H.expect
 '
 
+test_expect_success 'colored blame colors continuous fields' '
+
+	git mv hello.c world.c &&
+	git commit -a -m "moved file" &&
+	cat <<-EOF >> world.c &&
+	void world()
+	{
+		puts("world");
+	}
+	EOF
+	git add world.c &&
+	GIT_AUTHOR_NAME="F" GIT_AUTHOR_EMAIL="[email protected]" \
+		git commit -m "forgot to add changes to moved file" &&
+
+	git blame --abbrev=12 --color-fields world.c >actual.raw &&
+	test_decode_color <actual.raw >actual &&
+
+	grep "<BOLD;BLACK>hello.c" actual > colored_hello.expect &&
+	grep "hello.c" actual > all_hello.expect &&
+	test_line_count = 9 colored_hello.expect &&
+	test_line_count = 10 all_hello.expect &&
+
+	grep "<BOLD;BLACK>world.c" actual > colored_world.expect &&
+	grep "world.c" actual > all_world.expect &&
+	test_line_count = 3 colored_world.expect &&
+	test_line_count = 4 all_world.expect &&
+
+	grep "(F" actual > all_F.expect &&
+	grep "<BOLD;BLACK>(F" actual > colored_F.expect &&
+	test_line_count = 8 all_F.expect &&
+	test_line_count = 5 colored_F.expect &&
+
+	grep "(H" actual > all_H.expect &&
+	grep "<BOLD;BLACK>(H" actual > colored_H.expect &&
+	test_line_count = 5 all_H.expect &&
+	test_line_count = 3 colored_H.expect
+'
+
 test_done

Signature · 1 line

2.15.0.128.gcadd42da22

Stefan Beller

author

2017-11-10 01:10 UTC #4 lore raw

PATCH 4/4 builtin/blame: highlight recently changed lines

In reply to Stefan Beller

Choose a different color for dates and imitate a 'temperature cool down' for the dates.

Signed-off-by Stefan Beller <[email protected]>

---
 Documentation/config.txt | 20 +++++++++++++
 builtin/blame.c          | 74 +++++++++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 93 insertions(+), 1 deletion(-)

Documentation/config.txt +20 -0

index 85dfdc6a9b..7d2efc5ad6 100644
@@ -1194,6 +1194,26 @@ color.blame.repeatedMeta::
 	is repeated meta information per line (such as commit id,
 	author name, date and timezone). Defaults to dark gray.
 
+color.blame.highlightRecent::
+	This can be used to color the author and date of a blame line.
+	This overrides `color.blame.repeatedMeta` setting, which colors
+	repetitions.
++
+	This setting should be set to a comma-separated list of
+	color and date settings, starting and ending with a color, the dates
+	should be set from oldest to newest.
+	The metadata will be colored given the colors if the the line was
+	introduced before the given timestamp, overwriting older timestamped
+	colors.
++
+	Instead of an absolute timestamp relative timestamps work as
+	well, e.g. 2.weeks.ago is valid to address anything older than 2 weeks.
++
+	It defaults to "blue,12 month ago,white,1 month ago,red",
+	which colors everything older than one year blue, recent changes between
+	one month and one year old are kept white, and lines introduced within
+	the last month are colored red.
+
 color.ui::
 	This variable determines the default value for variables such
 	as `color.diff` and `color.grep` that control the use of color

builtin/blame.c +73 -1

134 lines not on this page · raw message

Signature · 1 line

2.15.0.128.gcadd42da22

Stefan Beller

author

2018-01-04 22:40 UTC #5 lore raw

v3 cover blame: (dim rep. metadata lines or fields, decay date coloring)

In reply to Stefan Beller

v3:

Thanks Eric for feedback, I implemented all of the suggestions. Specifically I dropped the abstractions in patch2 but keep around a similar abstraction in patch 3 as that still looks like it benefits (the condition is just growing large).

Thanks,
Stefan

v2:
This is picking up [1], but presenting it in another approach, as I realized these are orthogonal features:
* dimming repeated lines/fields of information
* giving a quick visual information how old (as a proxy for 'well tested') a line of code is.

Both features are configurable.

Changes from sending it out in November:
* better commit messages
* rebased on master

Any feedback welcome.

Thanks,
Stefan

[1] https://public-inbox.org/git/[email protected]/

Stefan Beller (4):
color.h: document and modernize header builtin/blame: dim uninteresting metadata builtin/blame: add option to color metadata fields separately builtin/blame: highlight recently changed lines

 Documentation/config.txt |  23 ++++++
 builtin/blame.c          | 201 ++++++++++++++++++++++++++++++++++++++++++-----
 color.c                  |   2 -
 color.h                  |  59 ++++++++++----
 t/t8012-blame-colors.sh  |  56 +++++++++++++
 5 files changed, 305 insertions(+), 36 deletions(-)
 create mode 100755 t/t8012-blame-colors.sh

Signature · 1 line

2.16.0.rc0.223.g4a4ac83678-goog

Stefan Beller

author

2018-01-04 22:40 UTC #6 lore raw

PATCH v3 1/4 color.h: document and modernize header

Add documentation explaining the functions in color.h. While at it, mark them extern.

Signed-off-by Stefan Beller <[email protected]>

---
 color.c |  2 --
 color.h | 58 ++++++++++++++++++++++++++++++++++++++++++++--------------
 2 files changed, 44 insertions(+), 16 deletions(-)

color.c +0 -2

index d48dd947c9..c4dc1cb989 100644
@@ -399,8 +399,6 @@ static int color_vfprintf(FILE *fp, const char *color, const char *fmt,
 	return r;
 }
 
-
-
 int color_fprintf(FILE *fp, const char *color, const char *fmt, ...)
 {
 	va_list args;

color.h +44 ~~-14~~

70 lines not on this page · raw message

Signature · 1 line

2.16.0.rc0.223.g4a4ac83678-goog

Junio C Hamano

2018-01-08 19:14 UTC #7 lore raw

Re: [PATCHv3 1/4] color.h: document and modernize header

color.h extern int color_stdout_is_tty; · line 82

Quoted 9 lines · Stefan Beller

/*
- * Set the color buffer (which must be COLOR_MAXLEN bytes)
- * to the raw color bytes; this is useful for initializing
- * default color variables.
+ * NEEDSWWORK: document this function or refactor grep.c to stop using this
+ * function.
*/
color.h +1 -1
-void color_set(char *dst, const char *color_bytes);
+extern void color_set(char *dst, const char *color_bytes);

The original that is removed by the patch documents the function well enough; as long as the NEEDSWORK comment is followed through in a later step in the series, it's alright, though ;-)

color.h extern int color_stdout_is_tty; · line 91

Quoted 12 lines

color.h +8 -4

-int git_config_colorbool(const char *var, const char *value);
-int want_color(int var);
-int color_parse(const char *value, char *dst);
-int color_parse_mem(const char *value, int len, char *dst);
+/*
+ * Parse a config option, which can be a boolean or one of
+ * "never", "auto", "always". Return a constant of
+ * GIT_COLOR_NEVER for "never" or negative boolean,
+ * GIT_COLOR_ALWAYS for "always" or a positive boolean,
+ * and GIT_COLOR_AUTO for "auto".
+ */
+extern int git_config_colorbool(const char *var, const char *value);

"never" and "always" not being part of usual boolean vocabulary makes it a bit awkward to explain.

Quoted 7 lines

+/*
+ * Output the formatted string in the specified color (and then reset to normal
+ * color so subsequent output is uncolored). Omits the color encapsulation if
+ * color is NULL. The color_fprintf_ln prints a new line after resetting
+ * the color. The color_print_strbuf prints the given pre-formatted strbuf
+ * instead, up to its first NUL character.
+ */

Obviously, it does not have to be part of this step nor series, but the above observation makes us realize that color_print_strbuf() would probably be an unreasonably narrow interface. It is not too much to ask the caller to dereference and pass only the .buf component of the strbuf to an alternative helper that takes "const char *" and by doing so would allow us to let other callers that do not have a strbuf but just a plain string use it, too.

Looks good.

Stefan Beller

author

2018-01-04 22:40 UTC #8 lore raw

PATCH v3 2/4 builtin/blame: dim uninteresting metadata

In reply to Stefan Beller

Signed-off-by Stefan Beller <[email protected]>

---
 Documentation/config.txt |  5 +++++
 builtin/blame.c          | 50 ++++++++++++++++++++++++++++++++++--------------
 color.h                  |  1 +
 t/t8012-blame-colors.sh  | 18 +++++++++++++++++
 4 files changed, 60 insertions(+), 14 deletions(-)
 create mode 100755 t/t8012-blame-colors.sh

Documentation/config.txt +5 -0

index 64c1dbba94..45749a574d 100644
@@ -1216,6 +1216,11 @@ color.status.<slot>::
 	status short-format), or
 	`unmerged` (files which have unmerged changes).
 
+color.blame.repeatedMeta::
+	Use the customized color for the part of git-blame output that
+	is repeated meta information per line (such as commit id,
+	author name, date and timezone). Defaults to dark gray.
+
 color.ui::
 	This variable determines the default value for variables such
 	as `color.diff` and `color.grep` that control the use of color

builtin/blame.c +36 ~~-14~~

118 lines not on this page · raw message

color.h +1 -0

index 2e768a10c6..2df2f86698 100644
@@ -30,6 +30,7 @@ struct strbuf;
 #define GIT_COLOR_BLUE		"\033[34m"
 #define GIT_COLOR_MAGENTA	"\033[35m"
 #define GIT_COLOR_CYAN		"\033[36m"
+#define GIT_COLOR_DARK		"\033[1;30m"
 #define GIT_COLOR_BOLD_RED	"\033[1;31m"
 #define GIT_COLOR_BOLD_GREEN	"\033[1;32m"
 #define GIT_COLOR_BOLD_YELLOW	"\033[1;33m"

t/t8012-blame-colors.sh new file +18 -0

new file mode 100755
index 0000000000..a2b7090cef
@@ -0,0 +1,18 @@
+#!/bin/sh
+
+test_description='colored git blame'
+. ./test-lib.sh
+
+PROG='git blame -c'
+. "$TEST_DIRECTORY"/annotate-tests.sh
+
+test_expect_success 'colored blame colors contiguous lines' '
+	git blame --abbrev=12 --color-lines hello.c >actual.raw &&
+	test_decode_color <actual.raw >actual &&
+	grep "<BOLD;BLACK>(F" actual > F.expect &&
+	grep "<BOLD;BLACK>(H" actual > H.expect &&
+	test_line_count = 2 F.expect &&
+	test_line_count = 3 H.expect
+'
+
+test_done

Signature · 1 line

2.16.0.rc0.223.g4a4ac83678-goog

Junio C Hamano

2018-01-08 19:34 UTC #9 lore raw

Re: [PATCHv3 2/4] builtin/blame: dim uninteresting metadata

Documentation/config.txt color.status.<slot>:: · line 1223

Quoted 6 lines · Stefan Beller

Documentation/config.txt +5 -0

+color.blame.repeatedMeta::
+	Use the customized color for the part of git-blame output that
+	is repeated meta information per line (such as commit id,
+	author name, date and timezone). Defaults to dark gray.
+
...

"Dark gray on default background" may alleviate worrries from those who prefer black ink on white paper display by hinting that both foreground and background colors can be configured.

Do we want to make this overridable from the command line, i.e. --color-repeated-meta=gray?

Quoted 1 line

+#define OUTPUT_COLOR_LINE 02000

The name of the macro implies that this is (or at least can be) a lot more generic UI request than merely "paint the same metadata on adjacent lines in a different color".

builtin/blame.c int cmd_blame(int argc, const char **argv, const char *prefix) · line 701

Quoted 1 line

builtin/blame.c +1 -0

+		OPT_BIT(0, "color-lines", &output_option, N_("color redundant metadata from previous line differently"), OUTPUT_COLOR_LINE),

Should this eventually become "--color=<yes,no,auto>" that is more usual and generic, possibly defaulting to "auto" in the future, I wonder?

color.h struct strbuf; · line 36

Quoted 12 lines

color.h +1 -0

index 2e768a10c6..2df2f86698 100644
@@ -30,6 +30,7 @@ struct strbuf;
 #define GIT_COLOR_BLUE		"\033[34m"
 #define GIT_COLOR_MAGENTA	"\033[35m"
 #define GIT_COLOR_CYAN		"\033[36m"
+#define GIT_COLOR_DARK		"\033[1;30m"
 #define GIT_COLOR_BOLD_RED	"\033[1;31m"
 #define GIT_COLOR_BOLD_GREEN	"\033[1;32m"
 #define GIT_COLOR_BOLD_YELLOW	"\033[1;33m"

How about using CYAN just like "diff" output uses it to paint the least interesting lines? That way we will keep the "uninteresting is cyan" consistency for the default settings without adding a new color to the palette.

Stefan Beller

author

2018-02-08 21:19 UTC #10 lore raw

Re: [PATCHv3 2/4] builtin/blame: dim uninteresting metadata

Quoted 22 lines · Junio C Hamano

Documentation/config.txt color.status.<slot>:: · line 1223
Quoted 6 lines · Stefan Beller
Documentation/config.txt +5 -0
+color.blame.repeatedMeta::
+     Use the customized color for the part of git-blame output that
+     is repeated meta information per line (such as commit id,
+     author name, date and timezone). Defaults to dark gray.
+
...
"Dark gray on default background" may alleviate worrries from those who prefer black ink on white paper display by hinting that both foreground and background colors can be configured.
Do we want to make this overridable from the command line, i.e. --color-repeated-meta=gray?
Quoted 1 line
+#define OUTPUT_COLOR_LINE 02000
The name of the macro implies that this is (or at least can be) a lot more generic UI request than merely "paint the same metadata on adjacent lines in a different color".

Quoted 5 lines

Quoted 1 line
+ OPT_BIT(0, "color-lines", &output_option, N_("color redundant metadata from previous line differently"), OUTPUT_COLOR_LINE),
Should this eventually become "--color=<yes,no,auto>" that is more usual and generic, possibly defaulting to "auto" in the future, I wonder?

This sounds like a good change to make, though what does "yes" mean here? The following patches introduce other modes of colorization, such that --color=<no, auto, lines, fields, decayed> would make sense to me, with "lines" as the mode of this patch, "fields" implemented in the next patch and "decayed" having the meaning of the heating algorithm presented in the last patch.

However each of these (except no and auto) would want to have extra parameters, which as mentioned above could go into its own separate parameters, such that

  git blame --color=line --color-repeated-meta=cyan

would seem like a good UI. This might look like it could be a good idea to have --color-repeated-meta imply --color=<line,auto> if no --color is given. But as different coloring modes will have different arguments to provide extra color information, I wonder what we would do with

  git blame --color="mode1" --extra-color-arg-for-mode-2

So I don't think the idea of having separate options would be a good idea as we'd have to think about the implications as mentioned above. So maybe

git blame "--color=lines,cyan yellow"

would work, if you want "cyan yellow" as the color configuration in the "lines" mode? (bad choice of a background color btw)

Quoted 9 lines

color.h +1 -0

index 2e768a10c6..2df2f86698 100644
@@ -30,6 +30,7 @@ struct strbuf;
 #define GIT_COLOR_BLUE               "\033[34m"
 #define GIT_COLOR_MAGENTA    "\033[35m"
 #define GIT_COLOR_CYAN               "\033[36m"
+#define GIT_COLOR_DARK               "\033[1;30m"

Quoted 8 lines

color.h struct strbuf; · line 36
Quoted 3 lines
color.h +0 -0
 #define GIT_COLOR_BOLD_RED   "\033[1;31m"
 #define GIT_COLOR_BOLD_GREEN "\033[1;32m"
 #define GIT_COLOR_BOLD_YELLOW        "\033[1;33m"
How about using CYAN just like "diff" output uses it to paint the least interesting lines? That way we will keep the "uninteresting is cyan" consistency for the default settings without adding a new color to the palette.

sounds good, I'll take that suggestion once we reach consensus on the UI for the user.

Thanks,
Stefan

Stefan Beller

author

2018-01-04 22:40 UTC #11 lore raw

PATCH v3 3/4 builtin/blame: add option to color metadata fields separately

In reply to Stefan Beller

Unlike the previous commit, this dims colors for each metadata field individually.

Signed-off-by Stefan Beller <[email protected]>

---
 builtin/blame.c         | 83 ++++++++++++++++++++++++++++++++++++++++++++-----
 t/t8012-blame-colors.sh | 38 ++++++++++++++++++++++
 2 files changed, 113 insertions(+), 8 deletions(-)

builtin/blame.c +75 -8

index 7b9c6e8676..60c8dadf4b 100644
@@ -323,6 +323,7 @@ static const char *format_time(timestamp_t time, const char *tz_str,
 #define OUTPUT_SHOW_EMAIL	0400
 #define OUTPUT_LINE_PORCELAIN 	01000
 #define OUTPUT_COLOR_LINE	02000
+#define OUTPUT_COLOR_FIELDS	04000
 
 static void emit_porcelain_details(struct blame_origin *suspect, int repeat)
 {

@@ -370,7 +371,56 @@ static void emit_porcelain(struct blame_scoreboard *sb, struct blame_entry *ent,
 		putchar('\n');
 }
 
-static void emit_other(struct blame_scoreboard *sb, struct blame_entry *ent, int opt)
+static int had_same_field_previously(int opt, int field,
+			  struct blame_entry *ent,
+			  struct blame_entry *prev)
+{
+	struct commit_info ci, prev_ci;
+
+	switch (field) {
+	case OUTPUT_SHOW_SCORE:
+		return ent->score == prev->score;
+	case OUTPUT_SHOW_NAME:
+		return prev->suspect &&
+			!strcmp(ent->suspect->path, prev->suspect->path);
+	case OUTPUT_SHOW_NUMBER:
+		return ent->s_lno == prev->s_lno + prev->num_lines - 1;
+
+	case OUTPUT_NO_AUTHOR:
+		get_commit_info(ent->suspect->commit, &ci, 1);
+		get_commit_info(prev->suspect->commit, &prev_ci, 1);
+		return ((opt & OUTPUT_SHOW_EMAIL) &&
+			!strcmp(ci.author_mail.buf, prev_ci.author_mail.buf)) ||
+			!strcmp(ci.author.buf, prev_ci.author.buf);
+	default:
+		BUG("unknown field");
+	}
+	return 0;
+}
+
+static void setup_field_color(int opt, int cnt, int field,
+			      struct blame_entry *ent,
+			      struct blame_entry *prev,
+			      const char **use_color,
+			      const char **reset_color)
+{
+	if (!(opt & OUTPUT_COLOR_FIELDS))
+		return;
+
+	if ((cnt > 0 ||
+	     (prev && had_same_field_previously(opt, field, ent, prev)))) {
+		*use_color = repeated_meta_color;
+		*reset_color = GIT_COLOR_RESET;
+	} else {
+		*use_color = "";
+		*reset_color = "";
+	}
+}
+
+static void emit_other(struct blame_scoreboard *sb,
+		       struct blame_entry *ent,
+		       struct blame_entry *prev,
+		       int opt)
 {
 	int cnt;
 	const char *cp;

@@ -414,18 +464,27 @@ static void emit_other(struct blame_scoreboard *sb, struct blame_entry *ent, int
 					   show_raw_time),
 			       ent->lno + 1 + cnt);
 		} else {
-			if (opt & OUTPUT_SHOW_SCORE)
+			if (opt & OUTPUT_SHOW_SCORE) {
+				setup_field_color(opt, cnt, OUTPUT_SHOW_SCORE,
+						  ent, prev, &color, &reset);
 				printf(" %s%*d %02d%s", color,
 				       max_score_digits, ent->score,
 				       ent->suspect->refcnt, reset);
-			if (opt & OUTPUT_SHOW_NAME)
+			}
+			if (opt & OUTPUT_SHOW_NAME) {
+				setup_field_color(opt, cnt, OUTPUT_SHOW_NAME,
+						   ent, prev, &color, &reset);
 				printf(" %s%-*.*s%s", color, longest_file,
 						      longest_file,
 						      suspect->path,
 						      reset);
-			if (opt & OUTPUT_SHOW_NUMBER)
+			}
+			if (opt & OUTPUT_SHOW_NUMBER) {
+				setup_field_color(opt, cnt, OUTPUT_SHOW_NUMBER,
+						  ent, prev, &color, &reset);
 				printf(" %s%*d%s", color, max_orig_digits,
 				       ent->s_lno + 1 + cnt, reset);
+			}
 			if (!(opt & OUTPUT_NO_AUTHOR)) {
 				const char *name;
 				int pad;

@@ -434,6 +493,8 @@ static void emit_other(struct blame_scoreboard *sb, struct blame_entry *ent, int
 				else
 					name = ci.author.buf;
 				pad = longest_author - utf8_strwidth(name);
+				setup_field_color(opt, cnt, OUTPUT_NO_AUTHOR,
+						  ent, prev, &color, &reset);
 				printf(" %s(%s%*s %10s%s", color,
 				       name, pad, "",
 				       format_time(ci.author_time,

@@ -459,7 +520,7 @@ static void emit_other(struct blame_scoreboard *sb, struct blame_entry *ent, int
 
 static void output(struct blame_scoreboard *sb, int option)
 {
-	struct blame_entry *ent;
+	struct blame_entry *ent, *prev = NULL;
 
 	if (option & OUTPUT_PORCELAIN) {
 		for (ent = sb->ent; ent; ent = ent->next) {

@@ -481,7 +542,8 @@ static void output(struct blame_scoreboard *sb, int option)
 		if (option & OUTPUT_PORCELAIN)
 			emit_porcelain(sb, ent, option);
 		else {
-			emit_other(sb, ent, option);
+			emit_other(sb, ent, prev, option);
+			prev = ent;
 		}
 	}
 }

@@ -699,6 +761,7 @@ int cmd_blame(int argc, const char **argv, const char *prefix)
 		OPT_BIT('e', "show-email", &output_option, N_("Show author email instead of name (Default: off)"), OUTPUT_SHOW_EMAIL),
 		OPT_BIT('w', NULL, &xdl_opts, N_("Ignore whitespace differences"), XDF_IGNORE_WHITESPACE),
 		OPT_BIT(0, "color-lines", &output_option, N_("color redundant metadata from previous line differently"), OUTPUT_COLOR_LINE),
+		OPT_BIT(0, "color-fields", &output_option, N_("color redundant metadata fields from previous line differently"), OUTPUT_COLOR_FIELDS),
 
 		/*
 		 * The following two options are parsed by parse_revision_opt()

@@ -757,6 +820,10 @@ int cmd_blame(int argc, const char **argv, const char *prefix)
 	revs.diffopt.flags.follow_renames = 0;
 	argc = parse_options_end(&ctx);
 
+	if ((output_option & OUTPUT_COLOR_LINE) &&
+	    (output_option & OUTPUT_COLOR_FIELDS))
+		die(_("cannot ask for colored lines and fields at the same time"));
+
 	if (incremental || (output_option & OUTPUT_PORCELAIN)) {
 		if (show_progress > 0)
 			die(_("--progress can't be used with --incremental or porcelain formats"));

@@ -960,8 +1027,8 @@ int cmd_blame(int argc, const char **argv, const char *prefix)
 
 	if (!(output_option & OUTPUT_PORCELAIN)) {
 		find_alignment(&sb, &output_option);
-		if ((output_option & OUTPUT_COLOR_LINE) &&
-		    !repeated_meta_color)
+		if (!repeated_meta_color &&
+		    (output_option & (OUTPUT_COLOR_LINE | OUTPUT_COLOR_FIELDS)))
 			repeated_meta_color = GIT_COLOR_DARK;
 	}

t/t8012-blame-colors.sh +38 -0

index a2b7090cef..18f9c9a16d 100755
@@ -15,4 +15,42 @@ test_expect_success 'colored blame colors contiguous lines' '
 	test_line_count = 3 H.expect
 '
 
+test_expect_success 'colored blame colors continuous fields' '
+
+	git mv hello.c world.c &&
+	git commit -a -m "moved file" &&
+	cat <<-EOF >> world.c &&
+	void world()
+	{
+		puts("world");
+	}
+	EOF
+	git add world.c &&
+	GIT_AUTHOR_NAME="F" GIT_AUTHOR_EMAIL="[email protected]" \
+		git commit -m "forgot to add changes to moved file" &&
+
+	git blame --abbrev=12 --color-fields world.c >actual.raw &&
+	test_decode_color <actual.raw >actual &&
+
+	grep "<BOLD;BLACK>hello.c" actual > colored_hello.expect &&
+	grep "hello.c" actual > all_hello.expect &&
+	test_line_count = 9 colored_hello.expect &&
+	test_line_count = 10 all_hello.expect &&
+
+	grep "<BOLD;BLACK>world.c" actual > colored_world.expect &&
+	grep "world.c" actual > all_world.expect &&
+	test_line_count = 3 colored_world.expect &&
+	test_line_count = 4 all_world.expect &&
+
+	grep "(F" actual > all_F.expect &&
+	grep "<BOLD;BLACK>(F" actual > colored_F.expect &&
+	test_line_count = 8 all_F.expect &&
+	test_line_count = 5 colored_F.expect &&
+
+	grep "(H" actual > all_H.expect &&
+	grep "<BOLD;BLACK>(H" actual > colored_H.expect &&
+	test_line_count = 5 all_H.expect &&
+	test_line_count = 3 colored_H.expect
+'
+
 test_done

Signature · 1 line

2.16.0.rc0.223.g4a4ac83678-goog

Stefan Beller

author

2018-01-04 22:40 UTC #12 lore raw

PATCH v3 4/4 builtin/blame: highlight recently changed lines

In reply to Stefan Beller

Choose a different color for dates and imitate a 'temperature cool down' depending upon age.

Originally I had planned to have the temperature cooldown dependent on the age of the project or file for example, as that might scale better, but that can be added on top of this commit, e.g. instead of giving a date, you could imagine giving a percentage that would be the linearly interpolated between now and the beginning of the file.

Signed-off-by Stefan Beller <[email protected]>

---
 Documentation/config.txt | 18 ++++++++++++
 builtin/blame.c          | 74 +++++++++++++++++++++++++++++++++++++++++++++++-
 2 files changed, 91 insertions(+), 1 deletion(-)

Documentation/config.txt +18 -0

index 45749a574d..aa967ce3ed 100644
@@ -1221,6 +1221,24 @@ color.blame.repeatedMeta::
 	is repeated meta information per line (such as commit id,
 	author name, date and timezone). Defaults to dark gray.
 
+color.blame.highlightRecent::
+	This can be used to color the author and date of a blame line.
+	This overrides `color.blame.repeatedMeta` setting, which colors
+	repetitions.
++
+This setting should be set to a comma-separated list of color and date settings,
+starting and ending with a color, the dates should be set from oldest to newest.
+The metadata will be colored given the colors if the the line was introduced
+before the given timestamp, overwriting older timestamped colors.
++
+Instead of an absolute timestamp relative timestamps work as well, e.g.
+2.weeks.ago is valid to address anything older than 2 weeks.
++
+It defaults to "blue,12 month ago,white,1 month ago,red", which colors
+everything older than one year blue, recent changes between one month and
+one year old are kept white, and lines introduced within the last month are
+colored red.
+
 color.ui::
 	This variable determines the default value for variables such
 	as `color.diff` and `color.grep` that control the use of color

builtin/blame.c +73 -1

134 lines not on this page · raw message

Signature · 1 line

2.16.0.rc0.223.g4a4ac83678-goog

Junio C Hamano

2018-01-08 19:56 UTC #13 lore raw

Re: [PATCHv3 4/4] builtin/blame: highlight recently changed lines

Quoted 14 lines · Stefan Beller

builtin/blame.c +14 -0

+static struct color_field {
+	timestamp_t hop;
+	char col[COLOR_MAXLEN];
+} *colorfield;
+static int colorfield_nr, colorfield_alloc;
+
+static void parse_color_fields(const char *s)
+{
+	struct string_list l = STRING_LIST_INIT_DUP;
+	struct string_list_item *item;
+	enum { EXPECT_DATE, EXPECT_COLOR } next = EXPECT_COLOR;
+
+	/* Ideally this would be stripped and split at the same time? */
+	string_list_split(&l, s, ',', -1);

builtin/blame.c static void setup_field_color(int opt, int cnt, int field, · line 445

Quoted 10 lines

builtin/blame.c +10 -0

+	ALLOC_GROW(colorfield, colorfield_nr + 1, colorfield_alloc);
+
+	for_each_string_list_item(item, &l) {
+		switch (next) {
+		case EXPECT_DATE:
+			colorfield[colorfield_nr].hop = approxidate(item->string);
+			next = EXPECT_COLOR;
+			colorfield_nr++;
+			ALLOC_GROW(colorfield, colorfield_nr + 1, colorfield_alloc);
+			break;

This should make sure cf[i].hop is monotonically increasing to avoid end-user mistakes, I would think (what's 'hop' by the way?).

builtin/blame.c static void setup_field_color(int opt, int cnt, int field, · line 448

Quoted 3 lines

builtin/blame.c +3 -0

+		case EXPECT_COLOR:
+			if (color_parse(item->string, colorfield[colorfield_nr].col))
+				die(_("expecting a color: %s"), item->string);

When you have a typo in one of your configuration files, say "[color "blame"] highlightrecent = 1,week,blue,...", you'd want to see a bit more than just "expecting a color: week" to help you diagnose and resolve the issue. Giving the name of the variable and the file the wrong definition was found in would be needed, givin that this is called from the config callback git_blame_config() below.

builtin/blame.c static void setup_field_color(int opt, int cnt, int field, · line 455

Quoted 7 lines

builtin/blame.c +7 -0

+			next = EXPECT_DATE;
+			break;
+		}
+	}
+
+	if (next == EXPECT_COLOR)
+		die (_("must end with a color"));

Same here.

builtin/blame.c int cmd_blame(int argc, const char **argv, const char *prefix) · line 835

Quoted 3 lines

builtin/blame.c +1 -0

 		OPT_BIT(0, "color-lines", &output_option, N_("color redundant metadata from previous line differently"), OUTPUT_COLOR_LINE),
 		OPT_BIT(0, "color-fields", &output_option, N_("color redundant metadata fields from previous line differently"), OUTPUT_COLOR_FIELDS),
+		OPT_BIT(0, "heated-lines", &output_option, N_("color lines by age"), OUTPUT_HEATED_LINES),

These options may be useful while having fun experimenting, but my gut feeling is that these are too fine-grained for end-users to tweak per invocation basis (which is what command line options are for).

But perhaps I am biased (as anybody else), as personally I find anything beyond step 2/4 uninteresting, and not adding too many of these options is consistent with that viewpoint ;-)

In any case, thanks for a fun read.

Stefan Beller

author

2018-01-08 20:09 UTC #14 lore raw

Re: [PATCHv3 4/4] builtin/blame: highlight recently changed lines

Quoted 55 lines · Junio C Hamano

Quoted 14 lines · Stefan Beller
builtin/blame.c +14 -0
+static struct color_field {
+     timestamp_t hop;
+     char col[COLOR_MAXLEN];
+} *colorfield;
+static int colorfield_nr, colorfield_alloc;
+
+static void parse_color_fields(const char *s)
+{
+     struct string_list l = STRING_LIST_INIT_DUP;
+     struct string_list_item *item;
+     enum { EXPECT_DATE, EXPECT_COLOR } next = EXPECT_COLOR;
+
+     /* Ideally this would be stripped and split at the same time? */
+     string_list_split(&l, s, ',', -1);
builtin/blame.c static void setup_field_color(int opt, int cnt, int field, · line 445
Quoted 10 lines
builtin/blame.c +10 -0
+     ALLOC_GROW(colorfield, colorfield_nr + 1, colorfield_alloc);
+
+     for_each_string_list_item(item, &l) {
+             switch (next) {
+             case EXPECT_DATE:
+                     colorfield[colorfield_nr].hop = approxidate(item->string);
+                     next = EXPECT_COLOR;
+                     colorfield_nr++;
+                     ALLOC_GROW(colorfield, colorfield_nr + 1, colorfield_alloc);
+                     break;
This should make sure cf[i].hop is monotonically increasing to avoid end-user mistakes, I would think (what's 'hop' by the way?).
builtin/blame.c static void setup_field_color(int opt, int cnt, int field, · line 448
Quoted 3 lines
builtin/blame.c +3 -0
+             case EXPECT_COLOR:
+                     if (color_parse(item->string, colorfield[colorfield_nr].col))
+                             die(_("expecting a color: %s"), item->string);
When you have a typo in one of your configuration files, say "[color "blame"] highlightrecent = 1,week,blue,...", you'd want to see a bit more than just "expecting a color: week" to help you diagnose and resolve the issue. Giving the name of the variable and the file the wrong definition was found in would be needed, givin that this is called from the config callback git_blame_config() below.
builtin/blame.c static void setup_field_color(int opt, int cnt, int field, · line 455
Quoted 7 lines
builtin/blame.c +7 -0
+                     next = EXPECT_DATE;
+                     break;
+             }
+     }
+
+     if (next == EXPECT_COLOR)
+             die (_("must end with a color"));
Same here.
Quoted 3 lines
OPT_BIT(0, "color-lines", &output_option, N_("color redundant metadata from previous line differently"), OUTPUT_COLOR_LINE), OPT_BIT(0, "color-fields", &output_option, N_("color redundant metadata fields from previous line differently"), OUTPUT_COLOR_FIELDS),
+ OPT_BIT(0, "heated-lines", &output_option, N_("color lines by age"), OUTPUT_HEATED_LINES),

Quoted 8 lines

These options may be useful while having fun experimenting, but my gut feeling is that these are too fine-grained for end-users to tweak per invocation basis (which is what command line options are for).
But perhaps I am biased (as anybody else), as personally I find anything beyond step 2/4 uninteresting, and not adding too many of these options is consistent with that viewpoint ;-)

See, I find 2 and 3 uninteresting and just did it 'because someone else hinted at that is what they want'. Maybe I was a bad listener.

4 (maybe with 2 in combination) would be all I need as that allows me to quickly judge the trustworthiness of code (old code is better, just like most liquors? ;)

Quoted 1 line

In any case, thanks for a fun read.

Thanks, I'll reread the comments and see if I can remove some options to make it useful for upstream consumption.

Thanks,
Stefan

Ævar Arnfjörð Bjarmason

2018-02-01 19:29 UTC #15 lore raw

In reply to Stefan Beller

On Thu, Jan 04 2018, Stefan Beller jotted:

Quoted 5 lines

Stefan Beller (4):
color.h: document and modernize header builtin/blame: dim uninteresting metadata builtin/blame: add option to color metadata fields separately builtin/blame: highlight recently changed lines

I like this feature in principle, but I can't get it to work. Building pu and:

    ./git -c color.ui=always --exec-path=$PWD blame Makefile

Shows no colors. Neither does:

    ./git -c color.ui=always --exec-path=$PWD -c color.blame.highlightRecent="red,12 month ago,blue" blame Makefile

And there's a bug, it segfaults on any custom value to the other config option:

    ./git -c color.ui=always --exec-path=$PWD -c color.blame.repeatedMeta=red blame Makefile

    0x00000000004c312b in color_parse_mem (value=0x8f6520 "red", value_len=3, dst=0x1 <Address 0x1 out of bounds>) at color.c:272
    272                     OUT('\033');

The repeated_meta_color variable is NULL when passed to color_parse_mem(). Didn't dig further.

Stefan Beller

author

2018-02-01 20:24 UTC #16 lore raw

Quoted 11 lines · Ævar Arnfjörð Bjarmason

On Thu, Jan 04 2018, Stefan Beller jotted:
Quoted 5 lines
Stefan Beller (4):
color.h: document and modernize header builtin/blame: dim uninteresting metadata builtin/blame: add option to color metadata fields separately builtin/blame: highlight recently changed lines
I like this feature in principle, but I can't get it to work. Building pu and:

Thanks for testing the feature!
Please use the flag --color-lines, --color-fields or --heated-lines for experimentation.

In the future we may decide that one of them becomes the default (which one?) and is triggered by the color.ui=always setting as well.

Quoted 9 lines

    ./git -c color.ui=always --exec-path=$PWD blame Makefile
Shows no colors. Neither does:
    ./git -c color.ui=always --exec-path=$PWD -c color.blame.highlightRecent="red,12 month ago,blue" blame Makefile
And there's a bug, it segfaults on any custom value to the other config option:

Quoted 7 lines

    ./git -c color.ui=always --exec-path=$PWD -c color.blame.repeatedMeta=red blame Makefile

    0x00000000004c312b in color_parse_mem (value=0x8f6520 "red", value_len=3, dst=0x1 <Address 0x1 out of bounds>) at color.c:272
    272                     OUT('\033');

The repeated_meta_color variable is NULL when passed to color_parse_mem(). Didn't dig further.

Thanks for noticing.
Fix below (white space mangled)

builtin/blame.c +1 -1

@@ -48,7 +48,7 @@ static int xdl_opts;
 static int abbrev = -1;
 static int no_whole_file_rename;
 static int show_progress;
-static char *repeated_meta_color;
+static char repeated_meta_color[COLOR_MAXLEN];
 
 static struct date_mode blame_date_mode = { DATE_ISO8601 };
 static size_t blame_date_width;

@@ -1099,9 +1099,9 @@ int cmd_blame(int argc, const char **argv, const

char *prefix)

        if (!(output_option & OUTPUT_PORCELAIN)) {
                find_alignment(&sb, &output_option);

- if (!repeated_meta_color &&
+ if (!*repeated_meta_color &&
(output_option & (OUTPUT_COLOR_LINE | OUTPUT_COLOR_FIELDS)))
- repeated_meta_color = GIT_COLOR_DARK;

+                       strcpy(repeated_meta_color, GIT_COLOR_DARK);
        }