Date: Thu, 29 Aug 2024 18:18:45 -0700
On Thursday 29 August 2024 17:23:02 GMT-7 Jeremy Rifkin via Std-Proposals
wrote:
> 1. People say it's portable/fine because every compiler supports it,
> but there's no documentation about exactly how it works and no source
> of truth for its semantics. 99.9% of people don't know what its
> potential pitfalls are.
> 2. There's a single source of truth for what it does, everyone can be
> on the same page about where it could break.
Ok so someone must do the work of collecting all the different behaviours and
then find out where it doesn't work. However:
1) the implementations differ, so standardising implies breaking the status
quo. That might mean that what works today may stop working.
2) there will be problems, which we know exist. So the recommendation should
be "this feature may not work for everyone, use at your own peril". Why should
we have that?
wrote:
> 1. People say it's portable/fine because every compiler supports it,
> but there's no documentation about exactly how it works and no source
> of truth for its semantics. 99.9% of people don't know what its
> potential pitfalls are.
> 2. There's a single source of truth for what it does, everyone can be
> on the same page about where it could break.
Ok so someone must do the work of collecting all the different behaviours and
then find out where it doesn't work. However:
1) the implementations differ, so standardising implies breaking the status
quo. That might mean that what works today may stop working.
2) there will be problems, which we know exist. So the recommendation should
be "this feature may not work for everyone, use at your own peril". Why should
we have that?
-- Thiago Macieira - thiago (AT) macieira.info - thiago (AT) kde.org Principal Engineer - Intel DCAI Platform & System Engineering
Received on 2024-08-30 01:18:48