LEWG-I wants first crack at this.

Cologne 2019-07 LEWGI Minutes

P1729R0 Text Parsing: Direction Review

Champion: Victor Zverovich

Minute Taker: Elias Kosunen

Start Overview: 07-15 11:00

We probably can't call this std::scan, because it clashes with std::inclusive_scan and std::exclusive_scan.

SG16 needs to review this.

Is the extension API essential for version 1?

Start Review: 11:16

Hana: How do you handle greediness?

Chris Kennelly: How much implementation experience do you have?

Adrian: Two things that this thing does: extract of strings, and conversion of strings to other types.

Start Polling: 11:27

POLL: We should promise more committee time to pursuing P1729R0, knowing that our time is scarce and this will leave less time for other work.

NO OBJECTION TO UNANIMOUS CONSENT.

Attendance: 15

End: 11:29

CONSENSUS: Bring a revision of P1729R0, exploring different approaches for outputs (e.g. output parameters, tuple-like return values), to LEWGI for further design review.

SG16 2019-07 review in Cologne

P1729R0

No polls taken. The author was present.

Dropping the SG16 label and looking forward to a future revision.

P1729R1 Text Parsing (Victor Zverovich , Elias Kosunen)

Belfast 2019-11 LEWGI Minutes

P1729R1 Text Parsing: Design Review

Champion: Victor Zverovich

Minute Taker: Ben Craig

Start Review: 11-04 13:44

Start Polling: 14:50

POLL: Using the same format string for input and output is an important design goal.

Attendance: 14

That has consensus.

POLL: We want positional arguments in std::scan.

Attendance: 14

That has consensus.

POLL: We want to require positional arguments to be unique in std::scan.

Attendance: 14

That has no consensus.

Break: 15:23

Resume: 15:45

POLL: Supporting non-DefaultConstructible types is an important design goal.

Attendance: 15

That has no consensus.

POLL: Getting the successfully scanned values from partial successes is an important design goal.

Attendance: 14

There is consensus against that.

End: 16:34

CONSENSUS: Bring a revision of P1729R1 (Text Parsing), with the guidance below, to LEWGI for further design review.

• Explore a simplified design that doesn't support partial success.
• Prioritize enabling the use of the same format string for input and output.
• Explore supporting non-unique positional arguments.

SG16 did not review P1729R1 in Belfast due to time constraints. We'll plan to review in Prague. The author has specifically requested guidance on how to specify and implement skipping of white space for Unicode encodings.

P1729R2 Text Parsing (Elias Kosunen, Victor Zverovich)

SG16 began reviewing P1729R2 during its 2023-09-27 meeting. No polls have been taken yet; continued review is planned for the 2023-10-11 SG16 meeting.

P1729R3 Text Parsing (Elias Kosunen)

SG16 reviewed a draft of P1729R3 during its 2023-10-11 meeting. We still haven't taken any polls, but the paper has strong support. There are still some questions to be sorted out regarding scanning of text vs binary input.

P1729R4 Text Parsing (Elias Kosunen, Victor Zverovich)

SG9 briefly looked at this paper in Tokyo to follow up on our discussion of the paper in Kona. We took the following poll:

With the following changes:

• make borrowed_tail_subrange_t exposition-only
• constrain the CharT of scannable_range to be exactly one of char or wchar_t

Forward P1729R4 to LEWG for inclusion in C++26.

Conclusion: enough of consensus (weak quorum), given the consensus that also existed in Kona
Attendance: 7

We will address the weak quorum by asking the SG9 mailing list to file any objections within the next week, but unless we get such objections (which will be summarized here), the SG9 aspects of this proposal are ready for LEWG review (once other study groups are done with it, of course).

SG16 reviewed P1729R4 during its 2024-03-13 meeting. No polls were taken.

Some questions were raised regarding the proposed std::scan() behavior for fill characters and centered text relative to the Python parse project, scanf(), and iostreams. These were raised while reviewing the examples in section 4.3.2, "Fill and align". The authors were encouraged to investigate further and return with a revised paper.

Adding the needs-revision label. SG16 looks forward to reviewing again in the near future.

P1729R5 Text Parsing (Elias Kosunen, Victor Zverovich)

@inbal2l, I don't think this paper is on track for C++26. If you agree, please remove the C++26 label.