Re: [cpp-netlib-devel] 0.8-devel progress, more async HTTP server handler goodness

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

In that case congrats are in order.

Congratulations. Great milestone there

Steve Obbayi

SKYPE: sobbayi
Software Developer

----- Original Message -----
From: "Dean Michael Berris" <mik...@gm...>
To: "C++ Networking Library Developers Mailing List" <cpp...@li...>
Sent: Wednesday, November 3, 2010 6:22:00 AM GMT -05:00 US/Canada Eastern
Subject: Re: [cpp-netlib-devel] 0.8-devel progress, more async HTTP server handler goodness

Hey Guys!

Aaaaaaaaaaand... I'm back! Earlier today was my daughter's first
birthday celebration, so that marks the end of my vacation. Hopefully
I'll get back to the hacking in the next couple days for the finishing
touches.

On Tue, Nov 2, 2010 at 5:07 AM, Glyn Matthews <gly...@gm...> wrote:
> Hi Dean,
>
> On 30 October 2010 17:09, Dean Michael Berris <mik...@gm...>
> wrote:
>>
>> 2. I'm close to finishing the implementation details of the
>> asynchronous HTTP connection. This is on target for the 1st week of
>> November deadline, so I'm confident we can ship the next release by
>> the end of that week.
>
> I get this error message when compiling:

Can you say what compiler you're using? Judging from the environment
and the error messages it looks like you're using GCC on Linux -- any
specific version number?

Also, I'm being bitten by things like these:

...
>mikhailberis/boost/network/protocol/http/algorithms/linearize.hpp:29:
> error: a call to a constructor cannot appear in a constant-expression

Which is not right because I don't use * in a constant-expression...
I'll look at that line again tomorrow.

...
> /home/gmatthew/Projects/cpp-netlib-mikhailberis/boost/network/protocol/http/server/async_connection.hpp:251:
> error: no matching function for call to ‘min(size_t&, long unsigned int)’
> /home/gmatthew/Projects/cpp-netlib-mikhailberis/boost/network/protocol/http/server/async_connection.hpp:267:
> error: no matching function for call to ‘min(size_t&, long unsigned int)’

Which looks like I just have to initialize a static local size_t to
hold the value of the macro. This type-specificity is actually really
hard to see -- but Clang seems to be able to rightly compile two types
of being `unsigned long int` (where std::size_t is also an unsigned
long int).

> /home/gmatthew/Projects/cpp-netlib-mikhailberis/boost/network/protocol/http/server/async_server.hpp:20:
> error: declaration of ‘typedef struct
> boost::network::http::response_header<Tag>
> boost::network::http::async_server_base<Tag, Handler>::response_header’
> /home/gmatthew/Projects/cpp-netlib-mikhailberis/boost/network/protocol/http/server/header/name.hpp:14:
> error: changes meaning of ‘response_header’ from ‘struct
> boost::network::http::response_header<Tag>’

This is also odd because response_header is a local typedef in the
async_server_base struct -- it doesn't change the meaning of the
earlier declared response_header template in the namespace scope. This
looks like a misinterpretation of the standard on the compiler's part
IMO.

It's easy to fix although it's perfectly standard C++ as far as I
understand it. I may be wrong though.

...
> /home/gmatthew/Projects/cpp-netlib-mikhailberis/boost/network/protocol/http/server/sync_connection.hpp:94:
> error:   in call to ‘find_if’
> /home/gmatthew/Projects/boost_1_43_0/boost/mpl/find_if.hpp:32: error:
> argument dependent lookup finds ‘template<class Sequence, class Predicate>
> struct boost::mpl::find_if’
> /home/gmatthew/Projects/cpp-netlib-mikhailberis/boost/network/protocol/http/server/sync_connection.hpp:94:
> error:   in call to ‘find_if’

What the... Why doesn't clang complain about this and does the right
thing? Let me check to see why this issue comes up. I'll try with GCC
4.5 and GCC 4.4.

> There are other open issues on github reporting compilation failures for
> other platforms too.
>

Yeah, thanks for those. I'll respond to Christian Henning on the
Boost-Users list about his issues.

>>
>> 4. There's already a skeleton/outline for the documentation of the
>> HTTP client and HTTP server interfaces. A reference manual is sorely
>> lacking in the documentation, and it would be good for others to get
>> their hands dirty with this one. I look forward to getting pull
>> requests for the documentation from those who would want to contribute
>> in this regard. :)
>
> Reference manual and examples :) we are still example poor at the moment.
>

Yeah, I agree. We really need more feedback on the documentation --
although people seem to be alright with looking at the examples and
running away with those, however poor the examples are. :D

Although the reference manual would be great to write, I'll need help
with making progress in that front. :)

>>
>> So that's it for now, I'll be in holiday mode starting tomorrow Oct 31
>> until Nov 2 (here in the Philippines). I'll be online briefly in the
>> morning Philippine time to check mail before I head on to my home
>> town, so if you have any questions you need answered before Wednesday
>> (PHT) please ask away so I can respond quickly. :)
>
> Have a relaxing holiday and I hope you don't pay attention to these mails
> before you return :)

Thanks Glyn! It's going to be hacking days for me starting tomorrow.
For now the email backlog is giving me the heebeejeebee's and I'm
longing for Inbox 0 again. :) I better get cracking on these emails
now. :D