From: Andrew Burgess <andrew.burgess@embecosm.com>
To: Tom Tromey <tom@tromey.com>
Cc: gdb-patches@sourceware.org
Subject: Re: [PATCH 1/2] gdb: don't print escape characters when a style is disabled
Date: Thu, 21 Jan 2021 16:08:08 +0000 [thread overview]
Message-ID: <20210121160808.GY265215@embecosm.com> (raw)
In-Reply-To: <20210121120618.GX265215@embecosm.com>
* Andrew Burgess <andrew.burgess@embecosm.com> [2021-01-21 12:06:18 +0000]:
> * Tom Tromey <tom@tromey.com> [2021-01-20 12:34:43 -0700]:
>
> > >>>>> "Andrew" == Andrew Burgess <andrew.burgess@embecosm.com> writes:
> >
> > Andrew> +# A replacement for the 'style' function found in gdb-utils.exp,
> > Andrew> +# filter out requests for the disabled style.
> > Andrew> +proc limited_style { str style } {
> >
> > Rather than rename the global 'style', I think it would be fine to just
> > have a wrapper proc that is used locally. That seems a lot less tricky.
> > Same with clean_restart.
> >
> > Andrew> /* Output pmatch with the highlight style. */
> > Andrew> - set_output_style (stream, highlight_style.style ());
> > Andrew> + if (!highlight_style.style ().is_default ())
> > Andrew> + set_output_style (stream, highlight_style.style ());
> >
> > set_output_style ends up setting applied_style.
> > So avoiding this call seems iffy to me.
>
> Tom,
>
> Thanks for the feedback.
>
> It's not clear to me if your feedback is about this one particular use
> of the 'if (!STYLE.is_default ()) { ... }' pattern, or if you are
> suggesting all uses of this pattern are suspect.
>
> It's just that you added the original uses of this pattern...
>
> So I guess my question would be, if its this one particular case then
> I don't understand why this cases is different to all the others,
> or...
>
> ... if it's all uses of this pattern then I guess my questions is,
> does this mean you think the original uses were a mistake? Or has
> something else changed ?
>
> > Maybe another approach would be to change emit_style_escape so that it
> > does nothing if the currently-applied style is equal to the passed-in
> > style?
>
> I'll give this a try and see if I can get this working.
OK, I think I have something working. There were some issues with
line wrapping, but I think that the final state is even better than
the original patch. Looking at some of the gdb.base/style.exp tests
the set of escape sequences emitted now seems to be minimal.
Let me know if this is what you were thinking of.
Thanks,
Andrew
---
commit c57874a4887009dee46687a895ea57157deb18b6
Author: Andrew Burgess <andrew.burgess@embecosm.com>
Date: Wed Jan 13 20:08:42 2021 +0000
gdb: don't print escape characters when a style is disabled
While working on another patch I noticed that if I disable a single
style with, for example:
set style filename background none
set style filename foreground none
set style filename intensity normal
Then in some places escape characters are still injected into the
output stream. This is a bit of an edge case, and I can't think when
this would actually cause problems, but it still felt like a bit of an
annoyance.
One place where this does impact is in testing, where it becomes
harder to write tight test patterns if it is not obvious when GDB will
decide to inject escape sequences.
It's especially annoying because depending on how something is printed
then GDB might, or might not, add escape characters. So this would
not add escape characters if the filename style was disabled:
fprintf_filtered (file, "%ps",
styled_string (file_name_style.style (),
"This is a test"));
But this would add escape characters:
fprintf_styled (file, file_name_style.style (),
"%s", "This is a test");
I tracked this down to some calls to set_output_style in utils.c.
Currently some calls to set_output_style (in utils.c) are guarded like
this:
if (!STYLE.is_default ())
set_output_style (stream, STYLE);
But some calls are not. It is the calls that are NOT guarded that
cause the extra escape sequences to be emitted.
My initial proposal to resolve this issue was simply to ensure that
all calls to set_output_style were guarded. The patch I posted for
this can be found here:
https://sourceware.org/pipermail/gdb-patches/2021-January/175096.html
The feedback on this proposal was that it might be better to guard
against the escape sequences being emitted at a later lever, right
down at emit_style_escape.
So this is what this version does. In emit_style_escape we already
track the currently applied style, so if the style we are being asked
to switch to is the same as the currently applied style then no escape
sequence needs to be emitted.
Making this change immediately exposed some issues in
fputs_maybe_filtered related to line wrapping. The best place to start
to understand what's going on with the styling and wrapping is look at
the test:
gdb.base/style.exp: all styles enabled: frame when width=20
If you run this test and then examine the output in an editor so the
escape sequences can be seen you'll see the duplicate escape sequences
that are emitted before this patch, the compare to after this patch
you'll see the set of escape sequences should be the minimum required.
In order to test these changes I have rewritten the gdb.base/style.exp
test script. The core of the script is now run multiple times. The
first time the test is run things are as they were before, all styles
are on.
After that the test is rerun multiple times. Each time through a
single style is disabled using the 3 explicit set calls listed above.
I then repeat all the tests, however, I arrange so that the patterns
for the disabled style now require no escape sequences.
gdb/ChangeLog:
* utils.c (emit_style_escape): Only emit an escape sequence if the
requested style is different than the current applied style.
(fputs_maybe_filtered): Adjust the juggling of the wrap_style, and
current applied_style.
(fputs_styled): Remove is_default check.
(fputs_styled_unfiltered): Likewise.
(vfprintf_styled_no_gdbfmt): Likewise.
gdb/testsuite/ChangeLog:
* gdb.base/style.exp (limited_style): New proc.
(clean_restart_and_disable): New proc.
(run_style_tests): New proc. Most of the old tests from this file
are now in this proc.
(test_startup_version_string): New proc. Reamining test from the
old file is in this proc.
diff --git a/gdb/testsuite/gdb.base/style.exp b/gdb/testsuite/gdb.base/style.exp
index aec1d0f42ed..08cd63546bd 100644
--- a/gdb/testsuite/gdb.base/style.exp
+++ b/gdb/testsuite/gdb.base/style.exp
@@ -17,203 +17,290 @@
standard_testfile
-save_vars { env(TERM) } {
- # We need an ANSI-capable terminal to get the output.
- setenv TERM ansi
-
- set test_macros 0
- set options debug
- get_compiler_info
- if { [test_compiler_info "gcc-*"] || [test_compiler_info "clang-*"] } {
- lappend options additional_flags=-g3
- set test_macros 1
- }
+# Compile the test executable.
+set test_macros 0
+set options debug
+get_compiler_info
+if { [test_compiler_info "gcc-*"] || [test_compiler_info "clang-*"] } {
+ lappend options additional_flags=-g3
+ set test_macros 1
+}
- if {[prepare_for_testing "failed to prepare" $testfile $srcfile \
- $options]} {
- return -1
- }
+if {[build_executable "failed to build" $testfile $srcfile $options]} {
+ return -1
+}
- set readnow [readnow]
+# The tests in this file are run multiple times with GDB's styles
+# disabled one at a time. This variable is the style that is
+# currently disabled.
+set currently_disabled_style ""
- if {![runto_main]} {
- fail "style tests failed"
- return
+# A wrapper around the 'style' function found in gdb-utils.exp,
+# filter out requests for the disabled style.
+proc limited_style { str style } {
+ global currently_disabled_style
+
+ if { $style != $currently_disabled_style } {
+ return [style $str $style]
}
- # Check that the source highlighter has not stripped away the leading
- # newlines.
- set main_line [gdb_get_line_number "break here"]
- gdb_test "list $main_line,$main_line" "return.*some_called_function.*"
+ return $str
+}
- gdb_test_no_output "set style enabled off"
+# A wrapper around 'clean_restart' from gdb.exp, this performs the
+# normal clean_restart, but then disables the currently disabled
+# style.
+proc clean_restart_and_disable { args } {
+ global currently_disabled_style
- set argv ""
- gdb_test_multiple "frame" "frame without styling" {
- -re -wrap "main \\(argc=.*, (argv=$hex)\\).*style\\.c:\[0-9\].*" {
- set argv $expect_out(1,string)
- pass $gdb_test_name
- }
- }
+ eval "clean_restart $args"
- gdb_test_no_output "set style enabled on"
+ if { $currently_disabled_style != "" } {
+ set st $currently_disabled_style
+ gdb_test_no_output "set style $st background none" ""
+ gdb_test_no_output "set style $st foreground none" ""
+ gdb_test_no_output "set style $st intensity normal" ""
+ }
+}
- set main_expr [style main function]
- set base_file_expr [style ".*style\\.c" file]
- set file_expr "$base_file_expr:\[0-9\]"
- set arg_expr [style "arg." variable]
+# The core of this test script. Run some tests of different aspects
+# of GDB's styling.
+#
+# Within this proc always use LIMITED_STYLE instead of STYLE, and
+# CLEAN_RESTART_AND_DISABLE instead of CLEAN_RESTART, this ensures
+# that the test operates as expected as styles are disabled.
+proc run_style_tests { } {
+ global testfile srcfile hex binfile test_macros
+ global currently_disabled_style decimal hex
+
+ save_vars { env(TERM) } {
+ # We need an ANSI-capable terminal to get the output.
+ setenv TERM ansi
+
+ # Restart GDB with the correct TERM variable setting, this
+ # means that GDB will enable styling.
+ clean_restart_and_disable ${binfile}
+
+ set readnow [readnow]
+
+ if {![runto_main]} {
+ fail "style tests failed"
+ return
+ }
- gdb_test "frame" \
- "$main_expr.*$arg_expr.*$arg_expr.*$file_expr.*"
- gdb_test "info breakpoints" "$main_expr at $file_expr.*"
+ # Check that the source highlighter has not stripped away the
+ # leading newlines.
+ set main_line [gdb_get_line_number "break here"]
+ gdb_test "list $main_line,$main_line" "return.*some_called_function.*"
- gdb_test_no_output "set style sources off"
- gdb_test "frame" \
- "\r\n\[^\033\]*break here.*" \
- "frame without sources styling"
- gdb_test_no_output "set style sources on"
+ gdb_test_no_output "set style enabled off"
- gdb_test "break -q main" "file $base_file_expr.*"
+ set argv ""
+ gdb_test_multiple "frame" "frame without styling" {
+ -re -wrap "main \\(argc=.*, (argv=$hex)\\).*style\\.c:\[0-9\].*" {
+ set argv $expect_out(1,string)
+ pass $gdb_test_name
+ }
+ }
- gdb_test "print &main" " = .* [style $hex address] <$main_expr>"
+ gdb_test_no_output "set style enabled on"
- # Regression test for a bug where line-wrapping would occur at the
- # wrong spot with styling. There were different bugs at different
- # widths, so try two.
- foreach width {20 30} {
- set argv_len [string length $argv]
- if { $argv_len == 0 } {
- continue
- }
+ set main_expr [limited_style main function]
+ set base_file_expr [limited_style ".*style\\.c" file]
+ set file_expr "$base_file_expr:\[0-9\]+"
+ set arg_expr [limited_style "arg." variable]
- # There was also a bug where the styling could be wrong in the
- # line listing; this is why the words from the source code are
- # spelled out in the final result line of the test.
- set re1_styled \
+ gdb_test "frame" \
[multi_line \
- "#0 *$main_expr.*$arg_expr.*" \
- ".*$arg_expr.*" \
- ".* at .*$file_expr.*" \
- "\[0-9\]+.*return.* break here .*"]
- set re2_styled \
- [multi_line \
- "#0 *$main_expr.*$arg_expr.*" \
- ".*$arg_expr.* at .*$file_expr.*" \
- "\[0-9\]+.*return.* break here .*"]
-
- # The length of the line containing argv containing:
- # - 4 leading spaces
- # - argv string
- # - closing parenthesis
- set line_len [expr 4 + $argv_len + 1]
-
- if { $line_len > $width } {
- # At on the next line.
- set re_styled $re1_styled
- } else {
- # At on the same line as argv.
- set re_styled $re2_styled
+ "#0\\s+$main_expr\\s+\\($arg_expr=$decimal,\\s+$arg_expr=$hex\\)\\s+at\\s+$file_expr" \
+ "\[0-9\]+\\s+.*return.* break here .*"]
+ gdb_test "info breakpoints" "$main_expr at $file_expr.*"
+
+ gdb_test_no_output "set style sources off"
+ gdb_test "frame" \
+ "\r\n\[^\033\]*break here.*" \
+ "frame without sources styling"
+ gdb_test_no_output "set style sources on"
+
+ gdb_test "break -q main" "file $base_file_expr.*"
+
+ gdb_test "print &main" " = .* [limited_style $hex address] <$main_expr>"
+
+ # Regression test for a bug where line-wrapping would occur at
+ # the wrong spot with styling. There were different bugs at
+ # different widths, so try two.
+ foreach width {20 30} {
+ set argv_len [string length $argv]
+ if { $argv_len == 0 } {
+ continue
+ }
+
+ # There was also a bug where the styling could be wrong in
+ # the line listing; this is why the words from the source
+ # code are spelled out in the final result line of the
+ # test.
+ set re1_styled \
+ [multi_line \
+ "#0\\s+$main_expr\\s+\\($arg_expr=$decimal,\\s+" \
+ "\\s+$arg_expr=$hex\\)" \
+ "\\s+at\\s+$file_expr" \
+ "\[0-9\]+\\s+.*return.* break here .*"]
+ set re2_styled \
+ [multi_line \
+ "#0\\s+$main_expr\\s+\\($arg_expr=.*" \
+ "\\s+$arg_expr=$hex\\)\\s+at\\s+$file_expr" \
+ "\[0-9\]+\\s+.*return.* break here .*"]
+
+ # The length of the line containing argv containing:
+ # - 4 leading spaces
+ # - argv string
+ # - closing parenthesis
+ set line_len [expr 4 + $argv_len + 1]
+
+ if { $line_len > $width } {
+ # At on the next line.
+ set re_styled $re1_styled
+ } else {
+ # At on the same line as argv.
+ set re_styled $re2_styled
+ }
+
+ gdb_test_no_output "set width $width"
+ gdb_test "frame" $re_styled "frame when width=$width"
}
- gdb_test_no_output "set width $width"
- gdb_test "frame" $re_styled "frame when width=$width"
- }
-
- # Reset width back to 0.
- gdb_test_no_output "set width 0"
+ # Reset width back to 0.
+ gdb_test_no_output "set width 0" ""
- if {$test_macros} {
- set macro_line [gdb_get_line_number "\#define SOME_MACRO"]
- gdb_test "info macro SOME_MACRO" \
- "Defined at $base_file_expr:$macro_line\r\n#define SOME_MACRO 23"
- }
+ if {$test_macros} {
+ set macro_line [gdb_get_line_number "\#define SOME_MACRO"]
+ gdb_test "info macro SOME_MACRO" \
+ "Defined at $base_file_expr:$macro_line\r\n#define SOME_MACRO 23"
+ }
- set main [style main function]
- set func [style some_called_function function]
- # Somewhere should see the call to the function.
- gdb_test "disassemble main" \
- [concat "Dump of assembler code for function $main:.*" \
- "[style $hex address].*$func.*"]
+ gdb_test_no_output "set width 0"
+
+ set main [limited_style main function]
+ set func [limited_style some_called_function function]
+ # Somewhere should see the call to the function.
+ gdb_test "disassemble main" \
+ [concat "Dump of assembler code for function $main:.*" \
+ "[limited_style $hex address].*$func.*"]
+
+ set ifield [limited_style int_field variable]
+ set sfield [limited_style string_field variable]
+ set efield [limited_style e_field variable]
+ set evalue [limited_style VALUE_TWO variable]
+ gdb_test "print struct_value" \
+ "\{$ifield = 23,.*$sfield = .*,.*$efield = $evalue.*"
+
+ set address_style_expr [limited_style ".*\".*address.*\".*style.*" address]
+ set color "blue"
+ if { $currently_disabled_style == "address" } {
+ set color "none"
+ }
+ gdb_test "show style address foreground" \
+ "The ${address_style_expr} foreground color is: ${color}" \
+ "style name and style word styled using its own style in show style"
- set ifield [style int_field variable]
- set sfield [style string_field variable]
- set efield [style e_field variable]
- set evalue [style VALUE_TWO variable]
- gdb_test "print struct_value" \
- "\{$ifield = 23,.*$sfield = .*,.*$efield = $evalue.*"
+ set aliases_expr [limited_style ".*aliases.*" title]
+ set breakpoints_expr [limited_style ".*breakpoints.*" title]
+ gdb_test "help" \
+ [multi_line \
+ "List of classes of commands:" \
+ "" \
+ "${aliases_expr} -- User-defined aliases of other commands\." \
+ "${breakpoints_expr} -- Making program stop at certain points\." \
+ ".*" \
+ ] \
+ "help classes of commands styled with title"
+
+ set taas_expr [limited_style ".*taas.*" title]
+ set tfaas_expr [limited_style ".*tfaas.*" title]
+ set cut_for_thre_expr [limited_style "cut for 'thre" highlight]
+ gdb_test "apropos -v cut for 'thre" \
+ [multi_line \
+ "" \
+ "${taas_expr}" \
+ "Apply a command to all .*" \
+ "Usage:.*" \
+ "short${cut_for_thre_expr}ad apply.*" \
+ "" \
+ "${tfaas_expr}" \
+ "Apply a command to all .*" \
+ "Usage:.*" \
+ "short${cut_for_thre_expr}ad apply.*" \
+ ]
+
+ clean_restart_and_disable
+
+ set quoted [string_to_regexp $binfile]
+ set pass_re "Reading symbols from [limited_style $quoted file]\.\.\."
+ if { $readnow } {
+ set pass_re \
+ [multi_line \
+ $pass_re \
+ "Expanding full symbols from [limited_style $quoted file]\.\.\."]
+ }
+ gdb_test "file $binfile" \
+ $pass_re \
+ "filename is styled when loading symbol file" \
+ "Are you sure you want to change the file.*" \
+ "y"
+
+ gdb_test "pwd" "Working directory [limited_style .*? file].*"
+
+ gdb_test_no_output "set print repeat 3"
+ gdb_test "print {0,0,0,0,0,0,0,0}" \
+ " = \\{0 [limited_style {<repeats.*8.*times>} metadata]\\}"
+
+ gdb_test "show logging file" \
+ "The current logfile is \"[limited_style .*? file]\"\\..*"
+
+ # Check warnings are styled by setting a rubbish data
+ # directory.
+ gdb_test "set data-directory Makefile" \
+ "warning: [limited_style .*? file] is not a directory\\..*"
+ gdb_test "show data-directory" \
+ "GDB's data directory is \"[limited_style .*? file]\"\\..*"
+
+ # Check that deprecation styles command names.
+ gdb_test_no_output "maintenance deprecate p \"new_p\"" \
+ "maintenance deprecate p \"new_p\" /1/"
+ gdb_test "p 5" \
+ "Warning: '[limited_style p title]', an alias for the command '[limited_style print title]', is deprecated.*Use '[limited_style new_p title]'.*" \
+ "p deprecated warning, with replacement"
+ }
+}
+# A separate test from the above as the styled text this checks can't
+# currently be disabled (the text is printed too early in GDB's
+# startup process).
+proc test_startup_version_string { } {
gdb_exit
gdb_spawn
+ # Deliberate use of base STYLE proc here as the style of the
+ # startup version string can't (currently) be controlled.
set vers [style "GNU gdb.*" "35;1"]
- gdb_test "" "${vers}.*" \
- "version is styled"
-
- set address_style_expr [style ".*\".*address.*\".*style.*" address]
- gdb_test "show style address foreground" \
- "The ${address_style_expr} foreground color is: blue" \
- "style name and style word styled using its own style in show style"
-
- set aliases_expr [style ".*aliases.*" title]
- set breakpoints_expr [style ".*breakpoints.*" title]
- gdb_test "help" \
- [multi_line \
- "List of classes of commands:" \
- "" \
- "${aliases_expr} -- User-defined aliases of other commands\." \
- "${breakpoints_expr} -- Making program stop at certain points\." \
- ".*" \
- ] \
- "help classes of commands styled with title"
-
- set taas_expr [style ".*taas.*" title]
- set tfaas_expr [style ".*tfaas.*" title]
- set cut_for_thre_expr [style "cut for 'thre" highlight]
- gdb_test "apropos -v cut for 'thre" \
- [multi_line \
- "" \
- "${taas_expr}" \
- "Apply a command to all .*" \
- "Usage:.*" \
- "short${cut_for_thre_expr}ad apply.*" \
- "" \
- "${tfaas_expr}" \
- "Apply a command to all .*" \
- "Usage:.*" \
- "short${cut_for_thre_expr}ad apply.*" \
- ]
-
- set quoted [string_to_regexp $binfile]
- set pass_re "Reading symbols from [style $quoted file]\.\.\."
- if { $readnow } {
- set pass_re \
- [multi_line \
- $pass_re \
- "Expanding full symbols from [style $quoted file]\.\.\."]
+ gdb_test "" "${vers}.*" "version is styled at startup"
+}
+
+
+# Run tests with all styles in their default state.
+with_test_prefix "all styles enabled" {
+ run_style_tests
+}
+
+# Now, for each style in turn. Disable that style only and run the
+# test again. Things in that style should NOT now be styled.
+foreach style { title file function highlight variable \
+ address metadata } {
+ set currently_disabled_style $style
+ with_test_prefix "disable style $style" {
+ run_style_tests
}
- gdb_test "file $binfile" \
- $pass_re \
- "filename is styled when loading symbol file"
-
- gdb_test "pwd" "Working directory [style .*? file].*"
-
- gdb_test_no_output "set print repeat 3"
- gdb_test "print {0,0,0,0,0,0,0,0}" \
- " = \\{0 [style {<repeats.*8.*times>} metadata]\\}"
-
- gdb_test "show logging file" \
- "The current logfile is \"[style .*? file]\"\\..*"
-
- # Check warnings are styled by setting a rubbish data directory.
- gdb_test "set data-directory Makefile" \
- "warning: [style .*? file] is not a directory\\..*"
- gdb_test "show data-directory" \
- "GDB's data directory is \"[style .*? file]\"\\..*"
-
- # Check that deprecation styles command names.
- gdb_test_no_output "maintenance deprecate p \"new_p\"" \
- "maintenance deprecate p \"new_p\" /1/"
- gdb_test "p 5" \
- "Warning: '[style p title]', an alias for the command '[style print title]', is deprecated.*Use '[style new_p title]'.*" \
- "p deprecated warning, with replacement"
}
+
+# Finally, check the styling of the version string during startup.
+test_startup_version_string
diff --git a/gdb/utils.c b/gdb/utils.c
index 414e7b153a0..b9f8997ec80 100644
--- a/gdb/utils.c
+++ b/gdb/utils.c
@@ -1426,12 +1426,15 @@ static void
emit_style_escape (const ui_file_style &style,
struct ui_file *stream = nullptr)
{
- applied_style = style;
+ if (applied_style != style)
+ {
+ applied_style = style;
- if (stream == nullptr)
- wrap_buffer.append (style.to_ansi ());
- else
- stream->puts (style.to_ansi ().c_str ());
+ if (stream == nullptr)
+ wrap_buffer.append (style.to_ansi ());
+ else
+ stream->puts (style.to_ansi ().c_str ());
+ }
}
/* Set the current output style. This will affect future uses of the
@@ -1800,14 +1803,20 @@ fputs_maybe_filtered (const char *linebuffer, struct ui_file *stream,
prompt is given; and to avoid emitting style
sequences in the middle of a run of text, we track
this as well. */
- ui_file_style save_style;
+ ui_file_style save_style = applied_style;
bool did_paginate = false;
chars_printed = 0;
lines_printed++;
if (wrap_column)
{
- save_style = wrap_style;
+ /* We are about to insert a newline at an historic
+ location in the WRAP_BUFFER. Before we do we want to
+ restore the default style. To know if we actually
+ need to insert an escape sequence we must restore the
+ current applied style to how it was at the WRAP_COLUMN
+ location. */
+ applied_style = wrap_style;
if (stream->can_emit_style_escape ())
emit_style_escape (ui_file_style (), stream);
/* If we aren't actually wrapping, don't output
@@ -1822,10 +1831,7 @@ fputs_maybe_filtered (const char *linebuffer, struct ui_file *stream,
stream->puts ("\n");
}
else
- {
- save_style = applied_style;
- flush_wrap_buffer (stream);
- }
+ flush_wrap_buffer (stream);
/* Possible new page. Note that
PAGINATION_DISABLED_FOR_COMMAND might be set during
@@ -1841,8 +1847,19 @@ fputs_maybe_filtered (const char *linebuffer, struct ui_file *stream,
if (wrap_column)
{
stream->puts (wrap_indent);
+
+ /* Having finished inserting the wrapping we should
+ restore the style as it was at the WRAP_COLUMN. */
if (stream->can_emit_style_escape ())
- emit_style_escape (save_style, stream);
+ emit_style_escape (wrap_style, stream);
+
+ /* The WRAP_BUFFER will still contain content, and that
+ content might set some alternative style. Restore
+ APPLIED_STYLE as it was before we started wrapping,
+ this reflects the current style for the last character
+ in WRAP_BUFFER. */
+ applied_style = save_style;
+
/* FIXME, this strlen is what prevents wrap_indent from
containing tabs. However, if we recurse to print it
and count its chars, we risk trouble if wrap_indent is
@@ -1895,16 +1912,9 @@ void
fputs_styled (const char *linebuffer, const ui_file_style &style,
struct ui_file *stream)
{
- /* This just makes it so we emit somewhat fewer escape
- sequences. */
- if (style.is_default ())
- fputs_maybe_filtered (linebuffer, stream, 1);
- else
- {
- set_output_style (stream, style);
- fputs_maybe_filtered (linebuffer, stream, 1);
- set_output_style (stream, ui_file_style ());
- }
+ set_output_style (stream, style);
+ fputs_maybe_filtered (linebuffer, stream, 1);
+ set_output_style (stream, ui_file_style ());
}
/* See utils.h. */
@@ -1913,16 +1923,9 @@ void
fputs_styled_unfiltered (const char *linebuffer, const ui_file_style &style,
struct ui_file *stream)
{
- /* This just makes it so we emit somewhat fewer escape
- sequences. */
- if (style.is_default ())
- fputs_maybe_filtered (linebuffer, stream, 0);
- else
- {
- set_output_style (stream, style);
- fputs_maybe_filtered (linebuffer, stream, 0);
- set_output_style (stream, ui_file_style ());
- }
+ set_output_style (stream, style);
+ fputs_maybe_filtered (linebuffer, stream, 0);
+ set_output_style (stream, ui_file_style ());
}
/* See utils.h. */
@@ -2222,11 +2225,9 @@ vfprintf_styled_no_gdbfmt (struct ui_file *stream, const ui_file_style &style,
std::string str = string_vprintf (format, args);
if (!str.empty ())
{
- if (!style.is_default ())
- set_output_style (stream, style);
+ set_output_style (stream, style);
fputs_maybe_filtered (str.c_str (), stream, filter);
- if (!style.is_default ())
- set_output_style (stream, ui_file_style ());
+ set_output_style (stream, ui_file_style ());
}
}
next prev parent reply other threads:[~2021-01-21 16:08 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-01-14 9:56 [PATCH 0/2] Style related fixes Andrew Burgess
2021-01-14 9:56 ` [PATCH 1/2] gdb: don't print escape characters when a style is disabled Andrew Burgess
2021-01-20 19:34 ` Tom Tromey
2021-01-21 12:06 ` Andrew Burgess
2021-01-21 16:08 ` Andrew Burgess [this message]
2021-01-22 15:26 ` Tom Tromey
2021-01-22 15:21 ` Tom Tromey
2021-01-14 9:56 ` [PATCH 2/2] gdb: add new version style Andrew Burgess
2021-01-14 10:35 ` Andrew Burgess
2021-01-14 11:15 ` Lancelot SIX via Gdb-patches
2021-01-20 19:39 ` Tom Tromey
2021-01-14 14:08 ` Eli Zaretskii via Gdb-patches
2021-01-20 19:39 ` Tom Tromey
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=20210121160808.GY265215@embecosm.com \
--to=andrew.burgess@embecosm.com \
--cc=gdb-patches@sourceware.org \
--cc=tom@tromey.com \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox