P1279 std::breakpoint

[wg21bot]

P1279R0 std::breakpoint (Isabella Muerte)

[jensmaurer]

Marked as "backlog" on San Diego EWG-I wiki.

[jfbastien]

EWG-I in Kona:
Do we want SG20 Education to see this?
SF F N A SA
4 0 4 2 1
Do we want SG15 Tooling to see this?
SF F N A SA
4 7 1 0 0
Do we want to see something like this again in EWGI?
SF F N A SA
4 4 3 1 0
(A vote: have to think more about technical issues, want to see in SG15 first, introduces semantics outside the language)

[jensmaurer]

SG20 in Kona:
"We encourage Isabella to explore this proposal further"
SF | F | N | A | SA
7 | 5 | 3 | 1 | 0

[brycelelbach]

Denver 2019-09 Tooling Minutes for P1279

P1279R0 std::breakpoint

Champion: Isabella Muerte

Minute Taker: Mark Zeren

Start Overview: 14:15

Start Polling: 14:40

POLL: Forward P1279, modified to specify the std::breakpoint function as implementation-defined with a (non-normative) note describing how debuggers should use it, for C++23.

Strongly For	Weakly For	Neutral	Weakly Against	Strongly Against
9	6	0	0	0

Attendance: 16

CONSENSUS: Unanimously in favor.

Michael Spencer appointed to help redraft.

Forward to EWG(I) + LEWGI + SG1

Questions for other groups:

• Needs to be bikeshedded.
• Which header?
• Should it be freestanding?

End: 14:47

CONSENSUS: The Tooling Study Group sends P1279R0 (std::breakpoint), with the guidance below, to EWGI/LEWGI/SG1 for C++23.

• Specify the std::breakpoint function as implementation-defined with a (non-normative) note describing how debuggers should use it.

[brycelelbach]

Belfast 2019-11 LEWGI Minutes

P1279R0 std::breakpoint: Design Guidance

Chair: Bryce Adelstein Lelbach

Champion: Bryce Adelstein Lelbach

Minute Taker: Mark Hoemmen

Start Review: 11-05 11:32

Start Polling: 11:35

POLL: Throwing exceptions when at a std::breakpoint in the debugger and catching them in the program is a use case we should support.

Strongly For	Weakly For	Neutral	Weakly Against	Strongly Against
0	2	1	5	6

Attendance: 16

There is consensus against that.

POLL: Remove noexcept from std::breakpoint.

Strongly For	Weakly For	Neutral	Weakly Against	Strongly Against
1	2	1	6	5

Attendance: 16

F: I think making it noexcept limits implementation freedom.

There is consensus against that.

POLL: Exploring making std::breakpoint a macro.

Strongly For	Weakly For	Neutral	Weakly Against	Strongly Against
3	2	5	4	1

Attendance: 16

This has no consensus.

POLL: Name of std::breakpoint (vote for all you can tolerate, vote as many times as you like).

Options	Votes
breakpoint	13
break_point	7
brake	4
do_not_commit	1
debug_break	10
pause	3
debug_breakpoint	11
trap	1
debug_callback	1
debug_trap	1

Attendance: 17

POLL: Name of std::breakpoint (vote once for your favorite).

Options	Votes
breakpoint	6
break_point	0
debug_break	2
debug_breakpoint	6

Attendance: 17

POLL: Name of std::breakpoint.

Strongly breakpoint	Weakly breakpoint	Neutral	Weakly debug_breakpoint	Strongly debug_breakpoint
5	3	1	3	6

Attendance: 18

The consensus is to call this either breakpoint or debug_breakpoint; author's choice.

End: 12:08

CONSENSUS: LEWGI sends P1279R0 (std::breakpoint), with the guidance below, to LEWG and EWGI.

• Add a section discussing existing practice/non-portable implementations (MSVC, GCC, etc).

[ogiroux]

SG1 sees no SG1 concerns with std::breakpoint:
SF F N A SA
11 6 0 0 0
Unanimous consent

[tituswinters]

EWGI/EWG are going to handle this first.

[cor3ntin]

The author informed me that she is no longer pursuing this paper.
If you are interested in pursuing this work further, reach out to her to get the paper's sources.

Closing for now