Re: [boost] [Boost-announce] [metaparse] Review period starts May 25th and ends June 7th

Subject: Re: [boost] [Boost-announce] [metaparse] Review period starts May 25th and ends June 7th
From: Edward Diener (eldiener_at_[hidden])
Date: 2015-06-02 14:05:25

This is my review of the Metaparse library:

The original metaparse library was represented as a library within 
mpllibs in the old Boost directory structure. The documentation of this 
library was entirely a tutorial of the library.

The tutorial was a very good general explanation of how to use the 
library. As such it showed the potential of the library and that the 
author of the library gave a great amount of thought to how the library 
should be used. I would rate the tutorial itself as excellent.

But a tutorial by itself is inadequate as documentation to a Boost 
library. Documentation needs a reference of all the classes, functions, 
data, macros of the library whether it is a library of meatfunctions or 
a library of run-time constructs. Furthermore some explanation of how 
the different parts of the library are organized and meant to be used 
within that organization is needed in documentation is necessary for me 
to understand a software library. So while I would rate the tutorial as 
excellent the documentation was inadequate.

Sometime during the review process the presentation of what was being 
reviewed completely changed. A new version of the library was presented 
using the current Boost directory structure, with a very full 
documentation set and the link to the tutorial documentation in the 
original version being reviewed was removed.

I am no doubt a bit stodgier than most programmers but this is not 
acceptable during a Boost review process. I believe the rule that what 
one reviews must remain absolutely unchanged during the period of a 
review has to be enforced. Changing everything during the review 
process, as has been done with this library, causes havoc and confusion 
for reviewers.

So while I welcome the need for such a library, have recognized the high 
quality of the library, and feel that the library would be extremely 
useful for Boost library developers among others in adding a well 
thought out parsing framework to the tools a template metaprogrammer 
would have in designing his own library my vote, given the irregularity 
of how this library was originally presented to Boost I vote NO for 
acceptance of this library into Boost.

While I realize that it may be seen as a PITA for the developer and 
review manager I would love to see another review of this library with 
the actual library being reviewed not changing under me and being either 
what was secondarily presented as the library or as an improvement of 
what was secondarily presented as the library. Personally, and without 
any animosity whatsoever to the library developer or the review manager, 
I don't think this review can be considered a valid one given what has 
happened during the review. But it is not my call to decide on this and 
I can only vote based on my annoyance that this has happened.