Re: [PFR][review] Review of PFR ends Today

From: Edward Diener (eldiener_at_[hidden])
Date: 2020-10-07 18:15:09

On 10/7/2020 1:33 PM, Benedek Thaler via Boost wrote:
> The review of the PFR: Precise and Flat tuple Representation (ex
> Precise and Flat Reflection, ex Magic Get, ex PODs Flat Reflection)
> ends Today.
> Do not hesitate to send late reviews, there's still time.

I vote to ACCEPT the library.

My suggestions:

1) While I appreciate the links in the docs I think that an explanation 
of the POD-like type with which the library can work should be added. 
Merely saying "aggregate initialization", with the link to the 
explanation on cppreference.com, does not really tell me precisely with 
what type of structure the library can work.

2) Whatever "loophole" is should be explained fully. An end-user is 
going to have little idea of whether his compiler in C++14 mode exploits 
the "loophole" or not, nor will he know whether the PFR library is 
figuring things out for him regarding the "loophole" unless it has a 
decent practical explanation. Just pointing to a C++ paper regarding the 
C++14 loophole does not seem enough to me.

3) I am not against the global operators since the doc tells me, in the 
topic "Three ways of getting operators" how I can turn them off or on 
depending how I want to use the library in a TU.

For the particular structure which the library can "convert" to a tuple 
for further functionality in situations in which working generically 
with a tuple is useful, I think the library is useful for programmers. 
This is my main reason for voting to accept the library. I did not do 
any extensive testing of the library, nor did I look at the internal 
code, but I am satisfied the library works as advertised in its basic 
functionality of struct -> tuple usage.

Although the usefulness of the library is limited, it is a tool for 
working with a C-like data structs, which I am sure many C++ programmers 
still use in their code. Thus my vote for acceptance.