C++ Logo


Advanced search

Re: Performance requirements for Unicode views/types/algorithms

From: Jens Maurer <jens.maurer_at_[hidden]>
Date: Thu, 2 Mar 2023 17:06:13 +0100
On 02/03/2023 16.50, Daniel Krügler wrote:
> Am Do., 2. März 2023 um 16:45 Uhr schrieb Jens Maurer via SG16
> <sg16_at_[hidden]>:
>> On 02/03/2023 16.28, Steve Downey via SG16 wrote:
>>> That ICU apparently has poor performance in this area indicates a little that many users are OK with it. But it does also sound like there are use cases for eager algorithms that could provide better performance. Probably something like the ones that JeanHeyd was proposing for C?
>>> Incidentally, should the charconv functions get `span`ed? Those char* make me itch a little? Or would that be an ABI pessimization?
>> <charconv> was intentionally designed to be a bare-bones interface
>> where other abstractions can be layered on top.
>> People have a number of ergonomics gripes about the interface,
>> so feel free to propose wrappers that are "nicer" to use,
>> if you feel you can make a bang-for-the-buck argument for a
>> facility somewhere between (say) strtoul and <charconv>.
>> Jens
> One proposal that springs to my mind would be
> https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2022/p2584r0.pdf

which is in need of addressing LEWG comments since August 2022,
per https://github.com/cplusplus/papers/issues/1243


Received on 2023-03-02 16:06:18