From: Gavin Lambert (boost_at_[hidden])
Date: 2020-10-05 01:16:26
On 5/10/2020 14:05, Peter Dimov wrote:
>> Having said that, I've personally never found a use-case for a
>> tuple-like get interface for anything, so perhaps I'm just not the
>> target audience for this library.
>
> You are surely familiar with the two classic examples, automatically
> generating a hash function by memberwise iteration, and automatically
> generating a serialization/deserialization function, by same.
Hash function might be a valid purpose, but I often find that there are
specific members that you want to exclude from equality/hash for various
reasons. And hash function implementation should be tightly coupled to
the structure definition anyway, so there's less reason to abstract it away.
I don't consider this library suitable for (de)serialization purposes
since it provides no access to the member name.