Subject: Re: [boost] [review][beast] Review of Beast starts today : July 1 - July 10
From: Peter Koch Larsen (peter.koch.larsen_at_[hidden])
Date: 2017-07-03 13:44:03
Just one perhaps uninformed remark. I have glanced over this thread as
I like to keep in touch with what happens in boost, but have no need
for a websocket interface.
To me, Vinnies library seems to be a well-designed and capable
library. But it seems to me that the library tries to accomplish two
things that coud be separated: parsing of http-headers and
transmission of http-headers. My guess is that - as Niall pointed out
- there is a large group of developers that could benefit from the
parsing part alone, and probably also a group that would perform their
own parsing and just use the transfer part. If this is so, Beast
should be split into two libraries.
Perhaps a fast way to accomplish that would be to have a small helper
library that could convert to/from asio-buffers to e.g.
std::vector<(w)char> and std::(w)string? Even if this caused
suboptimal performance, my guess is that this approach would have
appeal and not frighten to many users. We have had libraries in boost
with suboptimal performance before (lexical_cast is one such example)
where later revisions reduced or eliminated the performance loss.
On Mon, Jul 3, 2017 at 3:21 PM, Vinnie Falco via Boost
> On Mon, Jul 3, 2017 at 5:33 AM, Niall Douglas via Boost
> <boost_at_[hidden]> wrote:
> You've said a lot and I'd like some clarity, is your issue with
> beast::http::basic_parser and beast::serializer that it depends on
> <boost/asio/buffer.hpp>? Would your needs be satisfied if it did not
> have this dependence?
> Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk