C++ Logo

sg7

Advanced search

Re: [isocpp-lib-ext] SG7 Meeting P1240R2 part 2 on February 23th at 16:00 UTC

From: Matus Chochlik <chochlik_at_[hidden]>
Date: Tue, 22 Feb 2022 07:41:04 +0100
Hi SG7

On Tue, Feb 8, 2022 at 11:17 AM Hana Dusíková via Lib-Ext <
lib-ext_at_[hidden]> wrote:

> Hello wonderful SG7! (and LEWG + EWG) ❤️
>
> We will have a reflection meeting where we will continue discussion about
> P1240R2 (
> http://www.open-std.org/jtc1/sc22/wg21/docs/papers/2022/p1240r2.pdf)
>
> Daveed will continue with his presentation from last time.
>

If at all possible I'd like for us also to have a discussion about the
direction and maybe some subsequent actions.
I have made some slides comparing the "value-based" (VB, aka
"non-hana-consteval") vs. "type-based" (TB, I guess "hana-consteval")
reflection designs and also quite a lot of measurements of
compilation-times:

https://matus-chochlik.github.io/mirror/latex/value_vs_type_based.pdf

Some of the rationale for the current design is backed by claims that:
1) the TB design results in 2 orders of magnitude longer compile-times
compared to VB and even that the TB case exhibits exponential growth in
compile time while the VB case doesn't.
2) STL algorithms can be reused with the VB API (vector<meta::info>)
without any change.

The measurements don't show the former and the latter is true only for
cases that don't involve splicing.
As I have already mentioned; anecdotally 9 out of 10 use-cases that I have
implemented actually do require splicing.

I'd like to see if we have consensus on the current direction (ideally by
doing another vote on that).



>
> *⚡️ Meeting will start February 23th (Wednesday) at 16:00 UTC (11am EST,
> 8am PCT, 17:00 CET) over zoom:*
>
> https://iso.zoom.us/j/97957398677?pwd=OWRQSGtJbDhpUmN2bVBlZ0htN24xdz09
>
> 🕠 Expected duration for this meeting will be hour and half.
>

--Matus

Received on 2022-02-22 06:41:16