P3329 R0 Healing the C++ Filter View

[wg21bot]

P3329R0 Healing the C++ Filter View (Nicolai Josuttis)

[dhollman]

SG9 in Hagenberg

SG9 saw this paper in Hagenberg. We took the following polls:

---

At least some of the examples in P3329R0 are surpising to ordinary programmers and user-hostile, and we should do something to make less surprising alternatives feasible.

SF	F	N	A	SA
6	3	3	0	0

Consensus
Author: SF
Attendence: 12

---

We would welcome a paper that proposes take_if_view (name TBD) that fixes at least some of these problems.

SF	F	N	A	SA
3	4	2	1	0

Consensus
Author: SF
Attendence: 12

---

Such a follow-up paper should propose a view that is only an input_range

SF	F	N	A	SA
1	5	1	2	0

Consensus
Author: A
Attendence: 12

---

Such a follow-up paper should propose a view that is const-iterable

SF	F	N	A	SA
6	2	1	2	1

Very weak consensus
Author: SF
Attendence: 12
SA: See P3431R0

---

We would like to make some kind (design TBD) of breaking change to filter_view and other views with similar issues.

SF	F	N	A	SA
3	2	2	2	3

No consensus
Author: SF
Attendence: 12

---

[foonathan]

We discussed a follow-up paper proposing a fix (P3725 https://wiki.edg.com/pub/Wg21sofia2025/SG9/P3725R0_filterview4inputranges_250603.pdf) in SG9 in Sofia. Putting the notes here for now, as it does not have an issue yet: https://wiki.edg.com/bin/view/Wg21sofia2025/NotesSG9P3725R0 We forwarded that paper to LEWG.

We would like to add conditional const overload of begin/end to ranges::filter_view as proposed by P3725R0.

SF	F	N	A	SA
1	8	2	3	0

Attendance: 18
Author: SF

Strong consensus in favor.

We would like to add the input_filter range adaptor as proposed by P3725R0.

SF	F	N	A	SA
1	8	6	0	0

Attendance: 18
Author: SF

Strong consensus in favor.

Forward P3725R0 to LEWG for inclusion in C++29.

No objection to unanimous consent.