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

Subject: Re: [boost] [metaparse] Review period starts May 25th and ends June 7th - ongoing
From: Louis Dionne (ldionne.2_at_[hidden])
Date: 2015-06-04 23:07:28

Louis Dionne <ldionne.2 <at> gmail.com> writes:

> [...]
> >
> I am currently going through Metaparse's tutorial, and here are some questions
> and comments I have so far. I will be posting more questions by replying to
> this message if some come up as I progress through the tutorial. I will
> provide a proper review of the library when I'm done.

What follows is my official review of the Metaparse library
-----------------------------------------------------------
I vote YES for the acceptation of Metaparse into Boost.

> - What is your evaluation of the design?

I think the design is fine. Not an expert, but it looks like a classic
parser combinator library, except the author had to deal with the quirks
of C++ template metaprogramming. It is built using the well known and
proven concepts established by the MPL: metafunctions, metafunction classes
and type-level sequences, which is a safe bet.

That being said, I would be really really curious to try and implement
something similar with value-level syntax, similarly to the way Hana does
compile-time computations with a value-level syntax. I think we might be
able to achieve something with the same syntax as Spirit but for building
compile-time parsers too. However, this is just my own curiosity speaking
and it does not take anything away from Metaparse.

> - What is your evaluation of the implementation?

I have not looked at the implementation in depth, but I think the author
pretty much pushed C++03 to its limits. The only comment I would have would
be to provide a modernized version of the MPLLIBS_STRING macro when the
compiler supports it, because that is a huge bottleneck for using the library.

> - What is your evaluation of the documentation?

Despite the concerns I raised in my other messages, I think the tutorial is
very well written and is very instructive. I think it covers the necessary
material to get started and I think it manages to do so quickly enough.
The reference is simple but well organized and easy to read. There are
usage examples for every component I looked at, which is really useful.

> - What is your evaluation of the potential usefulness of the library?

I think the library can be useful, but it is obviously targeted towards
library writers rather than application developers. I think Metaparse plays
in the same field as Proto; they are "meta" libraries used to build other
libraries.

> - Did you try to use the library?
>     - With what compiler?
>     - Did you have any problems?

Yes. I cloned the mpllibs/ repository and was able to build it using the
CMake-based build system. I built/ran all the unit tests and the examples
with the following compilers:
    Clang           3.4.2, 3.5.2, 3.6.1, trunk
    GCC             5.1, trunk

Everything went fine, except for a link error due to Boost.Test with C++14
compilers, but I don't think that's Metaparse's fault. I was building against
Boost 1.58. The whole process was very easy; it took me under 5 minutes to
set it up. I did not try to use the Boost.Build system.

> - How much effort did you put into your evaluation? A glance? A quick
>   reading? In-depth study?

I read all the "Getting Started" part of the documentation, and I had a look
at the reference too. I looked a bit at the implementation itself, but not
for very long. I played with the final calculator example on my own machine
for about an hour or so. All in all, I would say I have spent +6h on this
review.

> - Are you knowledgeable about the problem domain?

I have used Spirit and other "parser generators" in the past, so I am
familiar to the domain of parsing. I have used Proto, MPL, Fusion and
I have written the MPL11 and Hana libraries, so I have a good knowledge
of the domain of metaprogramming.

Regards,
Louis