test failed on OSX latest build

[Matt Clark]

I looks simple to fix except that I'm not positive if the tests are wrong or the responses... I think it's the tests, but I would rather not mess anything up. Testing/Tempory files attached.

-Matt

PS Thanks so much for getting this building on OSX!

[Dean Michael Berris]

Thanks Matt,

On Mon, Sep 24, 2012 at 2:38 AM, Matt Clark <ryu.t...@gmail.com> wrote:
> I looks simple to fix except that I'm not positive if the tests are wrong or
> the responses... I think it's the tests, but I would rather not mess
> anything up. Testing/Tempory files attached.
>

I'll give this a go -- can you mention what compiler you're using and
what version you're building?

Unfortunately I can't take a look at it until later in my day (Sydney time).

Cheers

>
> -Matt
>
>
> PS Thanks so much for getting this building on OSX!
>

Pleasure's mine. :)

--
Dean Michael Berris
www.deanberris.com

[Matt Clark]

I'm using gcc 4.7.2 to build the latest master (c79b637). 

Also, I seem to be having issues with creating a simple test server, but that may be a different issue all together. I am very excited to get this working as I can see me and my company using it a lot and hopefully getting to contribute back as well as we run across things, but it'll all come down to what I can get working and what I have time to work on^^; I'm sure you guys understand!

Thanks again!

[Dean Michael Berris]

On Mon, Sep 24, 2012 at 5:13 PM, Matt Clark <ryu.t...@gmail.com> wrote:
> I'm using gcc 4.7.2 to build the latest master (c79b637).
>

Oh, right. That's going to be a problem.

> Also, I seem to be having issues with creating a simple test server, but
> that may be a different issue all together. I am very excited to get this
> working as I can see me and my company using it a lot and hopefully getting
> to contribute back as well as we run across things, but it'll all come down
> to what I can get working and what I have time to work on^^; I'm sure you
> guys understand!
>

Yes, this is a problem -- the server implementation is incomplete on
master. It's going to take a couple of weeks at least of me hacking
nights away on an HTTP server implementation that meets the
requirements documented in this doc:

https://docs.google.com/document/d/1jUJxs99bcbxAFIWn4Slm8JIMMmrtKc3jwhBz6Da05Lk/edit

I would be very happy to have your help if you can comment on the
direction and whether you'd be able to contribute sooner than later in
the form of testing and/or feedback.

If you can wait that long or spare some time in helping me out
implementing it, that would be greatly appreciated.

Cheers

> Thanks again!
>
> On Sunday, September 23, 2012 6:47:36 PM UTC-5, Dean Michael Berris wrote:
>>
>> Thanks Matt,
>>
>> On Mon, Sep 24, 2012 at 2:38 AM, Matt Clark <ryu.t...@gmail.com> wrote:
>> > I looks simple to fix except that I'm not positive if the tests are
>> > wrong or
>> > the responses... I think it's the tests, but I would rather not mess
>> > anything up. Testing/Tempory files attached.
>> >
>>
>> I'll give this a go -- can you mention what compiler you're using and
>> what version you're building?
>>
>> Unfortunately I can't take a look at it until later in my day (Sydney
>> time).
>>
>> Cheers
>>
>> >
>> > -Matt
>> >
>> >
>> > PS Thanks so much for getting this building on OSX!
>> >
>>
>> Pleasure's mine. :)
>>
>> --
>> Dean Michael Berris
>> www.deanberris.com
>

> --
> You received this message because you are subscribed to the Google Groups
> "The C++ Network Library" group.
> To view this discussion on the web visit
> https://groups.google.com/d/msg/cpp-netlib/-/N8oKIx1grvQJ.
>
> To post to this group, send email to cpp-n...@googlegroups.com.
> To unsubscribe from this group, send email to
> cpp-netlib+...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/cpp-netlib?hl=en.



--
Dean Michael Berris | Software Engineer
Google

[Matt Clark]

Sorry to double post, but I don't see a way to edit my previous post. If you are interested in looking at my current test server code it's hosted here: https://github.com/mclark4386/cpp-netlib-test 

It's pretty much the example code slightly updated for some changes I could figure out and getting a response back. Also, the sample server example segfaults on me on response...

For my code when I try to compile I get:

/Projects/TESTS/Boost/cpp-netlib-test/main.cpp:19:19: error: incomplete type 'boost::network::http::response' named in nested name specifier

(sorry didn't mean to hijack this thread either>.<;)

[Matt Clark]

On Monday, September 24, 2012 2:16:07 AM UTC-5, Dean Michael Berris wrote:

On Mon, Sep 24, 2012 at 5:13 PM, Matt Clark <ryu.t...@gmail.com> wrote:
> I'm using gcc 4.7.2 to build the latest master (c79b637).
>

Oh, right. That's going to be a problem.

 

I tried clang as well (latest that the latest xcode installs and the version just before it), but it kept freaking out at the std::move calls etc. so I just build it with the gcc I use for c++11 coding. Perhaps I missed something along the way...

 

> Also, I seem to be having issues with creating a simple test server, but
> that may be a different issue all together. I am very excited to get this
> working as I can see me and my company using it a lot and hopefully getting
> to contribute back as well as we run across things, but it'll all come down
> to what I can get working and what I have time to work on^^; I'm sure you
> guys understand!
>

Yes, this is a problem -- the server implementation is incomplete on
master. It's going to take a couple of weeks at least of me hacking
nights away on an HTTP server implementation that meets the
requirements documented in this doc:

  https://docs.google.com/document/d/1jUJxs99bcbxAFIWn4Slm8JIMMmrtKc3jwhBz6Da05Lk/edit

I would be very happy to have your help if you can comment on the
direction and whether you'd be able to contribute sooner than later in
the form of testing and/or feedback.

If you can wait that long or spare some time in helping me out
implementing it, that would be greatly appreciated.

Cheers

I actually read through that doc right after I posted this. It looks like you are taking some hints from node.js and the like, so coming from those on the WS dev side of things I approve^^. 

I would love to help in anyway I can, but I feel I must warn you that my wife is probably going to give birth to our first child within the week. That will most likely be both good and bad for me testing/coding because it'll free me up from my day-to-day work for the most part and I'll have time when she is sleeping to code, but there is no telling how long her labor is going to take or how intense it will be so I may only have time to sleep^^;  However, C++ is/was my first love language-wise so finding a library for it that may make it reasonable to use it for WS dev is very exciting for me, so just let me know how I can help and I'm going to do everything I can within reason (without pissing off my wife lol).

[Dean Michael Berris]

On Tue, Sep 25, 2012 at 12:36 PM, Matt Clark <ryu.t...@gmail.com> wrote:
> On Monday, September 24, 2012 2:16:07 AM UTC-5, Dean Michael Berris wrote:
>>
>> On Mon, Sep 24, 2012 at 5:13 PM, Matt Clark <ryu.t...@gmail.com> wrote:
>> > I'm using gcc 4.7.2 to build the latest master (c79b637).
>> >
>>
>> Oh, right. That's going to be a problem.
>
>
> I tried clang as well (latest that the latest xcode installs and the version
> just before it), but it kept freaking out at the std::move calls etc. so I
> just build it with the gcc I use for c++11 coding. Perhaps I missed
> something along the way...
>

Ah, right. There's an undocumented way of doing this with Clang 3.1 and higher:

When building with CMake:

cmake -DCMAKE_CXX_COMPILER=clang++
-DCMAKE_CXX_COMPILE_FLAGS=-std=c++11 -DCMAKE_CXX_LINK_FLAGS=-std=c++11

If you want to use libc++ instead of libstdc++ (recommended actually on OS X):

cmake -DCMAKE_CXX_COMPILER=clang++
-DCMAKE_CXX_COMPILE_FLAGS="-std=c++1 -stdlib=libc++"
-DCMAKE_CXX_LINK_FLAGS="-std=c++11 -stdlib=libc++"

Not entirely sure whether it's CMAKE_CXX_COMPILE_FLAGS or
CMAKE_CXX_COMPILER_FLAGS -- play around locally to find out the
correct combination. :D

>
>>
>>
>> > Also, I seem to be having issues with creating a simple test server, but
>> > that may be a different issue all together. I am very excited to get
>> > this
>> > working as I can see me and my company using it a lot and hopefully
>> > getting
>> > to contribute back as well as we run across things, but it'll all come
>> > down
>> > to what I can get working and what I have time to work on^^; I'm sure
>> > you
>> > guys understand!
>> >
>>
>> Yes, this is a problem -- the server implementation is incomplete on
>> master. It's going to take a couple of weeks at least of me hacking
>> nights away on an HTTP server implementation that meets the
>> requirements documented in this doc:
>>
>>
>> https://docs.google.com/document/d/1jUJxs99bcbxAFIWn4Slm8JIMMmrtKc3jwhBz6Da05Lk/edit
>>
>> I would be very happy to have your help if you can comment on the
>> direction and whether you'd be able to contribute sooner than later in
>> the form of testing and/or feedback.
>>
>> If you can wait that long or spare some time in helping me out
>> implementing it, that would be greatly appreciated.
>>
>> Cheers
>
>
> I actually read through that doc right after I posted this. It looks like
> you are taking some hints from node.js and the like, so coming from those on
> the WS dev side of things I approve^^.
>

Yes, this is more idiomatic for server-side development. :)

> I would love to help in anyway I can, but I feel I must warn you that my
> wife is probably going to give birth to our first child within the week.
> That will most likely be both good and bad for me testing/coding because
> it'll free me up from my day-to-day work for the most part and I'll have
> time when she is sleeping to code, but there is no telling how long her
> labor is going to take or how intense it will be so I may only have time to
> sleep^^; However, C++ is/was my first love language-wise so finding a
> library for it that may make it reasonable to use it for WS dev is very
> exciting for me, so just let me know how I can help and I'm going to do
> everything I can within reason (without pissing off my wife lol).
>

Best wishes to you and your wife on your upcoming child! I understand
how stressfully exciting this time is -- no pressure though on coding,
family > code. I've been in exactly the same situation too so I can
empathize. :)

I'm definitely looking forward to building greater things together soon! :D

Cheers

[Matt Clark]

On Monday, September 24, 2012 10:53:57 PM UTC-5, Dean Michael Berris wrote:

On Tue, Sep 25, 2012 at 12:36 PM, Matt Clark <ryu.t...@gmail.com> wrote:
> On Monday, September 24, 2012 2:16:07 AM UTC-5, Dean Michael Berris wrote:
>>
>> On Mon, Sep 24, 2012 at 5:13 PM, Matt Clark <ryu.t...@gmail.com> wrote:
>> > I'm using gcc 4.7.2 to build the latest master (c79b637).
>> >
>>
>> Oh, right. That's going to be a problem.
>
>
> I tried clang as well (latest that the latest xcode installs and the version
> just before it), but it kept freaking out at the std::move calls etc. so I
> just build it with the gcc I use for c++11 coding. Perhaps I missed
> something along the way...
>

Ah, right. There's an undocumented way of doing this with Clang 3.1 and higher:

When building with CMake:

  cmake -DCMAKE_CXX_COMPILER=clang++
-DCMAKE_CXX_COMPILE_FLAGS=-std=c++11 -DCMAKE_CXX_LINK_FLAGS=-std=c++11

If you want to use libc++ instead of libstdc++ (recommended actually on OS X):

  cmake -DCMAKE_CXX_COMPILER=clang++
-DCMAKE_CXX_COMPILE_FLAGS="-std=c++1 -stdlib=libc++"
-DCMAKE_CXX_LINK_FLAGS="-std=c++11 -stdlib=libc++"

Not entirely sure whether it's CMAKE_CXX_COMPILE_FLAGS or
CMAKE_CXX_COMPILER_FLAGS -- play around locally to find out the
correct combination. :D

I add this into the compiler flags check in the CMakeLists.txt

elseif(${CMAKE_CXX_COMPILER_ID} MATCHES Clang)

  set(CMAKE_CXX_FLAGS "-std=c++11 -stdlib=libc++")

  set(CMAKE_CXX_LINK_FLAGS "-std=c++11 -stdlib=libc++")

  message("C++ Flags: ${CMAKE_CXX_FLAGS} link flags: ${CMAKE_CXX_LINK_FLAGS}")

but now it's erring on 

error: static_cast from '__node_base *' (aka '__list_node_base<value_type, __void_pointer> *') to '__node_pointer'

      (aka 'std::__1::__list_node<boost::spirit::info, void *> *') is not allowed

I run up on this before but I can't remember how it was fixed... or maybe it was just on clang.

Ditto! Should I just keep keeping an eye on the official master branch or is there a branch or fork that I should test with? I have my own fork, but not sure where things are and don't want to duplicate work or anything like that.^^

[Matt Clark]

forgot this is where I found the flags variable: http://www.cmake.org/Wiki/CMake_Useful_Variables

[Dean Michael Berris]

Cool!

> but now it's erring on
> error: static_cast from '__node_base *' (aka '__list_node_base<value_type,
> __void_pointer> *') to '__node_pointer'
> (aka 'std::__1::__list_node<boost::spirit::info, void *> *') is not
> allowed
>

Interesting. What version of Boost are you using again? I've mostly
been testing against Boost.Trunk and so far I've been able to build
things properly (though I've been in major refactor/overhaul mode).

> I run up on this before but I can't remember how it was fixed... or maybe it
> was just on clang.
>

Make sure you're using clang 3.1 or the "top of trunk". Also, make
sure that your libc++ implementation is not broken. :)

The official master branch is the active development branch. Releases
will have branches and will be supported for up to 1 year of release.

[Matt Clark]

It's not much and you may have already added it to your code, but I can do a pull request if you like.

 

> but now it's erring on
> error: static_cast from '__node_base *' (aka '__list_node_base<value_type,
> __void_pointer> *') to '__node_pointer'
>       (aka 'std::__1::__list_node<boost::spirit::info, void *> *') is not
> allowed
>

Interesting. What version of Boost are you using again? I've mostly
been testing against Boost.Trunk and so far I've been able to build
things properly (though I've been in major refactor/overhaul mode).

I am not positive to be honest. > 1.50. I just downloaded it from the site. I think I'll grab the latest and try again with that.

 

> I run up on this before but I can't remember how it was fixed... or maybe it
> was just on clang.
>

Make sure you're using clang 3.1 or the "top of trunk". Also, make
sure that your libc++ implementation is not broken. :)

clang -v:

Apple clang version 4.1 (tags/Apple/clang-421.11.65) (based on LLVM 3.1svn)

Target: x86_64-apple-darwin12.2.0

Thread model: posix

It's the version that XCode 4.5 installs. 

Awesome! I'll just keep watching that!

[Dean Michael Berris]

Please send in a pull request -- that would be most appreciated!

>>
>>
>> > but now it's erring on
>> > error: static_cast from '__node_base *' (aka
>> > '__list_node_base<value_type,
>> > __void_pointer> *') to '__node_pointer'
>> > (aka 'std::__1::__list_node<boost::spirit::info, void *> *') is
>> > not
>> > allowed
>> >
>>
>> Interesting. What version of Boost are you using again? I've mostly
>> been testing against Boost.Trunk and so far I've been able to build
>> things properly (though I've been in major refactor/overhaul mode).
>
>
> I am not positive to be honest. > 1.50. I just downloaded it from the site.
> I think I'll grab the latest and try again with that.
>

Interesting. I haven't seen this before.

>>
>>
>> > I run up on this before but I can't remember how it was fixed... or
>> > maybe it
>> > was just on clang.
>> >
>>
>> Make sure you're using clang 3.1 or the "top of trunk". Also, make
>> sure that your libc++ implementation is not broken. :)
>
>
> clang -v:
> Apple clang version 4.1 (tags/Apple/clang-421.11.65) (based on LLVM 3.1svn)
> Target: x86_64-apple-darwin12.2.0
> Thread model: posix
>
> It's the version that XCode 4.5 installs.
>

Ah, right. So it looks like you're using LLVM 3.1 (where Clang 3.1
should be based from). Apple may have their own modifications there,
but I don't know enough about the difference between Clang's trunk and
Apple's released compiler. I should really be testing on the released
compiler too but... that's for some other day. :D

Thanks Matt!

[Matt Clark]

Done!

 

>>
>>
>> > but now it's erring on
>> > error: static_cast from '__node_base *' (aka
>> > '__list_node_base<value_type,
>> > __void_pointer> *') to '__node_pointer'
>> >       (aka 'std::__1::__list_node<boost::spirit::info, void *> *') is
>> > not
>> > allowed
>> >
>>
>> Interesting. What version of Boost are you using again? I've mostly
>> been testing against Boost.Trunk and so far I've been able to build
>> things properly (though I've been in major refactor/overhaul mode).
>
>
> I am not positive to be honest. > 1.50. I just downloaded it from the site.
> I think I'll grab the latest and try again with that.
>

Interesting. I haven't seen this before.

Upon further investigation it seems I installed it with macports. Also, it is 1.50. I have pulled the latest trunk and am working on installing it now.

 

>>
>>
>> > I run up on this before but I can't remember how it was fixed... or
>> > maybe it
>> > was just on clang.
>> >
>>
>> Make sure you're using clang 3.1 or the "top of trunk". Also, make
>> sure that your libc++ implementation is not broken. :)
>
>
> clang -v:
> Apple clang version 4.1 (tags/Apple/clang-421.11.65) (based on LLVM 3.1svn)
> Target: x86_64-apple-darwin12.2.0
> Thread model: posix
>
> It's the version that XCode 4.5 installs.
>

Ah, right. So it looks like you're using LLVM 3.1 (where Clang 3.1
should be based from). Apple may have their own modifications there,
but I don't know enough about the difference between Clang's trunk and
Apple's released compiler. I should really be testing on the released
compiler too but... that's for some other day. :D

lol, but isn't that what I'm for =P

[Dean Michael Berris]

On Wed, Sep 26, 2012 at 11:41 AM, Matt Clark <ryu.t...@gmail.com> wrote:
> On Tuesday, September 25, 2012 8:29:06 PM UTC-5, Dean Michael Berris wrote:
>>
>> Please send in a pull request -- that would be most appreciated!
>
>
> Done!
>

Sweet -- let me a have a look and merge it. :)

>>
>>
>> >>
>> >>
>> >> > but now it's erring on
>> >> > error: static_cast from '__node_base *' (aka
>> >> > '__list_node_base<value_type,
>> >> > __void_pointer> *') to '__node_pointer'
>> >> > (aka 'std::__1::__list_node<boost::spirit::info, void *> *') is
>> >> > not
>> >> > allowed
>> >> >
>> >>
>> >> Interesting. What version of Boost are you using again? I've mostly
>> >> been testing against Boost.Trunk and so far I've been able to build
>> >> things properly (though I've been in major refactor/overhaul mode).
>> >
>> >
>> > I am not positive to be honest. > 1.50. I just downloaded it from the
>> > site.
>> > I think I'll grab the latest and try again with that.
>> >
>>
>> Interesting. I haven't seen this before.
>
>
> Upon further investigation it seems I installed it with macports. Also, it
> is 1.50. I have pulled the latest trunk and am working on installing it now.
>

I remember issues with 1.50 and boost.move interfering with C++11 move.

>>
>>
>> >>
>> >>
>> >> > I run up on this before but I can't remember how it was fixed... or
>> >> > maybe it
>> >> > was just on clang.
>> >> >
>> >>
>> >> Make sure you're using clang 3.1 or the "top of trunk". Also, make
>> >> sure that your libc++ implementation is not broken. :)
>> >
>> >
>> > clang -v:
>> > Apple clang version 4.1 (tags/Apple/clang-421.11.65) (based on LLVM
>> > 3.1svn)
>> > Target: x86_64-apple-darwin12.2.0
>> > Thread model: posix
>> >
>> > It's the version that XCode 4.5 installs.
>> >
>>
>> Ah, right. So it looks like you're using LLVM 3.1 (where Clang 3.1
>> should be based from). Apple may have their own modifications there,
>> but I don't know enough about the difference between Clang's trunk and
>> Apple's released compiler. I should really be testing on the released
>> compiler too but... that's for some other day. :D
>
>
> lol, but isn't that what I'm for =P
>

Ha! Indeed. :)

Thanks again Matt!

Cheers

[Matt Clark]

On Tuesday, September 25, 2012 8:54:41 PM UTC-5, Dean Michael Berris wrote:

On Wed, Sep 26, 2012 at 11:41 AM, Matt Clark <ryu.t...@gmail.com> wrote:
> On Tuesday, September 25, 2012 8:29:06 PM UTC-5, Dean Michael Berris wrote:
>> >> > but now it's erring on
>> >> > error: static_cast from '__node_base *' (aka
>> >> > '__list_node_base<value_type,
>> >> > __void_pointer> *') to '__node_pointer'
>> >> >       (aka 'std::__1::__list_node<boost::spirit::info, void *> *') is
>> >> > not
>> >> > allowed
>> >> >
>> >>
>> >> Interesting. What version of Boost are you using again? I've mostly
>> >> been testing against Boost.Trunk and so far I've been able to build
>> >> things properly (though I've been in major refactor/overhaul mode).
>> >
>> >
>> > I am not positive to be honest. > 1.50. I just downloaded it from the
>> > site.
>> > I think I'll grab the latest and try again with that.
>> >
>>
>> Interesting. I haven't seen this before.
>
>
> Upon further investigation it seems I installed it with macports. Also, it
> is 1.50. I have pulled the latest trunk and am working on installing it now.
>

I remember issues with 1.50 and boost.move interfering with C++11 move.

I may try 1.51. The trunk is erring out when I try to compile cpp-netlib on:

/opt/local/include/boost/chrono/duration.hpp:369:44: error: constexpr function never produces a constant expression

        static BOOST_CONSTEXPR long double lowest() throw()

                                           ^

/opt/local/include/boost/chrono/duration.hpp:371:21: note: non-constexpr function 'max' cannot be used in a constant expression

            return -(std::numeric_limits<long double>::max)();

                    ^

/usr/bin/../lib/c++/v1/limits:443:43: note: declared here

    _LIBCPP_INLINE_VISIBILITY static type max() _NOEXCEPT {return __base::max();}

which looks to me like it's choking on libc++ because libc++ doesn't make max a constexpr... could be mis-reading it though.

Honestly not sure who to report that to lol... so I'm just going to try it with 1.51 and hope that it behaves better with libc++.

 

>>
>>
>> >>
>> >>
>> >> > I run up on this before but I can't remember how it was fixed... or
>> >> > maybe it
>> >> > was just on clang.
>> >> >
>> >>
>> >> Make sure you're using clang 3.1 or the "top of trunk". Also, make
>> >> sure that your libc++ implementation is not broken. :)
>> >
>> >
>> > clang -v:
>> > Apple clang version 4.1 (tags/Apple/clang-421.11.65) (based on LLVM
>> > 3.1svn)
>> > Target: x86_64-apple-darwin12.2.0
>> > Thread model: posix
>> >
>> > It's the version that XCode 4.5 installs.
>> >
>>
>> Ah, right. So it looks like you're using LLVM 3.1 (where Clang 3.1
>> should be based from). Apple may have their own modifications there,
>> but I don't know enough about the difference between Clang's trunk and
>> Apple's released compiler. I should really be testing on the released
>> compiler too but... that's for some other day. :D
>
>
> lol, but isn't that what I'm for =P
>

Ha! Indeed. :)

Thanks again Matt!

Not a problem, just hoping I can be more helpful then annoying lol.

[Matt Clark]

got the same error with 1.51 as with 1.50. Clean all the boost libs and includes just to make sure nothing was corrupting the install for trunk and got the same constexpr error.  ... so I'm going to play with it and see what I can do^^;

[Dean Michael Berris]

Hmmm... either this is a GCC bug that Boost is relying on, or there's
something wrong with the Apple released version of libc++ -- what
version of OS X are you on? I'm building with Mountain Lion BTW. This
may mean you have to do the scary thing of installing libc++ or not
relying on libc++ (libstdc++ should be OK on OS X IIRC).

Can you give that a shot?

[Matt Clark]

I'm on Mountain Lion as well. I'll see about getting the dev versions of clang, etc and just be careful in installing them^^ Anytime that I try to use clang without libc++ I get the std::move errors.

[Marshall Clow]

On Sep 25, 2012, at 9:06 PM, Matt Clark <ryu.t...@gmail.com> wrote:
> I'm on Mountain Lion as well. I'll see about getting the dev versions of clang, etc and just be careful in installing them^^ Anytime that I try to use clang without libc++ I get the std::move errors.

Yes; if you try to use c++11 w/o libc++, you'll get those kinds of errors.

I wrote a post about that
http://marshall.calepin.co/llvmclang-and-standard-libraries-on-mac-os-x.html

-- Marshall

Marshall Clow Idio Software <mailto:mclow...@gmail.com>

A.D. 1517: Martin Luther nails his 95 Theses to the church door and is promptly moderated down to (-1, Flamebait).
-- Yu Suzuki

[Matt Clark]

Ah, that makes a lot of sense. Thanks for the explanation!

Dean, I'm going to get the latest versions of clang and libc++ and I'll let you know how it turns out when I get to it^^; (Lost of contractions today... very busy doing those last minute things)

[Matt Clark]

Dean, had you already install the CLI build tools from xcode before you installed your clang and libcxx?

[Dean Michael Berris]

Yes, although I think the problem here is a little subtle -- you may
need to rebuild Boost using Clang and libc++ too to make this go away.
That said I really cannot track this down at the moment -- still
deeply involved in something not remotely related to this -- but if
anybody else has experience with this I'd defer to them and their
experience.

Marshall may have more ideas on this as I believe he hacks on Boost
with clang on OS X.

On Thu, Sep 27, 2012 at 12:14 PM, Matt Clark <ryu.t...@gmail.com> wrote:
> Dean, had you already install the CLI build tools from xcode before you
> installed your clang and libcxx?
>

> --
> You received this message because you are subscribed to the Google Groups
> "The C++ Network Library" group.
> To view this discussion on the web visit

> https://groups.google.com/d/msg/cpp-netlib/-/0T5QCTEddncJ.

>
> To post to this group, send email to cpp-n...@googlegroups.com.
> To unsubscribe from this group, send email to
> cpp-netlib+...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/cpp-netlib?hl=en.

[Matt Clark]

I actually think I'll have it with just the edge libc++ if I can just get it to link to it... sorry didn't mean to bug you. Just trying to get this compiling as quickly as possible so that I can start testing/helping ^^; Hopefully I'll get it tonight^^

[Dean Michael Berris]

On Thu, Sep 27, 2012 at 1:44 PM, Matt Clark <ryu.t...@gmail.com> wrote:
> I actually think I'll have it with just the edge libc++ if I can just get it
> to link to it... sorry didn't mean to bug you. Just trying to get this
> compiling as quickly as possible so that I can start testing/helping ^^;
> Hopefully I'll get it tonight^^
>

Sounds good to me. I'll look forward to you being able to start
building things with Clang on Mac OS X.

Cheers

[Marshall Clow]

On Sep 26, 2012, at 8:16 PM, Dean Michael Berris <dbe...@google.com> wrote:

> Yes, although I think the problem here is a little subtle -- you may
> need to rebuild Boost using Clang and libc++ too to make this go away.
> That said I really cannot track this down at the moment -- still
> deeply involved in something not remotely related to this -- but if
> anybody else has experience with this I'd defer to them and their
> experience.
>
> Marshall may have more ideas on this as I believe he hacks on Boost
> with clang on OS X.

I would try with the released clang/libc++ (part of Xcode 4.5) before I started building them.

-- Marshall

P.S. I'll see what I can find out tomorrow.

>
> On Thu, Sep 27, 2012 at 12:14 PM, Matt Clark <ryu.t...@gmail.com> wrote:
>> Dean, had you already install the CLI build tools from xcode before you
>> installed your clang and libcxx?
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "The C++ Network Library" group.
>> To view this discussion on the web visit
>> https://groups.google.com/d/msg/cpp-netlib/-/0T5QCTEddncJ.
>>
>> To post to this group, send email to cpp-n...@googlegroups.com.
>> To unsubscribe from this group, send email to
>> cpp-netlib+...@googlegroups.com.
>> For more options, visit this group at
>> http://groups.google.com/group/cpp-netlib?hl=en.
>
>
>
> --
> Dean Michael Berris | Software Engineer
> Google
>
> --
> You received this message because you are subscribed to the Google Groups "The C++ Network Library" group.
> To post to this group, send email to cpp-n...@googlegroups.com.
> To unsubscribe from this group, send email to cpp-netlib+...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/cpp-netlib?hl=en.
>

[Dean Michael Berris]

Thanks Marshall!

I've dug into this a little and I can see that the problem is libc++
that comes with Mountain Lion doesn't have numeric_limits<...>::max()
marked as constexpr. This I believe is by design as it seems the
compiler that shipped in xcode 4.5 is already capable of constexpr,
but the libc++ implementation installed on the base system doesn't
have numeric_limits<...>::max() defined as constexpr. Because Boost is
enabled constexpr using Apple's version of clang installed in the base
system, this leads to Boost.Chrono having issues with the
implementation.

I don't have this problem with a suitably new version of Clang and
libc++ installed at an alternate location. I would think that this
should be raised to the Boost developers (but I don't have time nor
patience right now to do that).

If anything I would consider having the latest Clang and libc++ build
installed at an alternate location instead.

Cheers

[Matt Clark]

Oooook, so I have some good new and some bad news.

Good news:

-I have the lib compiling...

Bad news:

-I had to install a VM of linux to do so :-/ 

-2 tests fail in the make test

I don't have much time right now, but I'll debug it later and see what's up. Just wanted to shot you guys an update. Hopefully I'll be able to start helping soon!^^

On Sunday, September 23, 2012 11:38:04 AM UTC-5, Matt Clark wrote:

I looks simple to fix except that I'm not positive if the tests are wrong or the responses... I think it's the tests, but I would rather not mess anything up. Testing/Tempory files attached.

[Dean Michael Berris]

On Sun, Oct 14, 2012 at 2:58 PM, Matt Clark <ryu.t...@gmail.com> wrote:
> Oooook, so I have some good new and some bad news.
>
> Good news:
> -I have the lib compiling...
>

Cool!

> Bad news:
> -I had to install a VM of linux to do so :-/

Hmmm... that's not good.

> -2 tests fail in the make test
>

I don't see these on the Mac though. Is this ToT Clang+libc++?

> I don't have much time right now, but I'll debug it later and see what's up.
> Just wanted to shot you guys an update. Hopefully I'll be able to start
> helping soon!^^
>

Awesome, thanks Matt!

Cheers

[Matt Clark]

On Monday, October 15, 2012 4:18:30 PM UTC-5, Dean Michael Berris wrote:

On Sun, Oct 14, 2012 at 2:58 PM, Matt Clark <ryu.t...@gmail.com> wrote:
> Oooook, so I have some good new and some bad news.
>
> Good news:
> -I have the lib compiling...
>

Cool!

> Bad news:
> -I had to install a VM of linux to do so :-/

Hmmm... that's not good.

> -2 tests fail in the make test
>

I don't see these on the Mac though. Is this ToT Clang+libc++?

TBH I don't know that it has anything to do with that. I think it has more to do with where you live/how your network is set up(at least for one of the failed tests), but I could be wrong.

Here's why I think that:

Here is the jist of the log from the test:

15/21 Testing: cpp-netlib-http-client_get_test

Running 2 test cases...

/cpp-netlib/libs/network/test/http/client_get_test.cpp:29: error in "http_client_get_test": check status_ == 302u failed [200 != 302]

/cpp-netlib/libs/network/test/http/client_get_test.cpp:30: error in "http_client_get_test": check status_message_ == std::string("Found") failed [OK != Found]

/cpp-netlib/libs/network/test/http/client_get_test.cpp:50: error in "https_client_get_test": check status_ == 302u failed [200 != 302]

/cpp-netlib/libs/network/test/http/client_get_test.cpp:51: error in "https_client_get_test": check status_message_ == std::string("Found") failed [OK != Found] 

Now in the code for that test 302u and std::string("Found") are both hard coded. So I'm getting a straight up OK 200 status from the connection to "http://www.google.com" which is a good thing^^ It seems you are used to getting a redirecting 302 Found with your setup, most likely because you aren't in the US and are therefore getting redirected to the regional version of google^^ 

I could be wrong, but I think that this may be better stated as BOOST_CHECK(status_==302u||status_==200) and BOOST_CHECK(status_message_ == std::string("Found")||status_message_ == std::string("OK")) .

What do you think?

I'll be happy to test it do a pull request if you would like^^

Cheers,

-Matt

PS Just got the news today that if we haven't had our child by Tuesday the 23rd the doctors will be inducing the birth^^

[Dean Michael Berris]

On Mon, Oct 15, 2012 at 7:52 PM, Matt Clark <ryu.t...@gmail.com> wrote:
>
>
> On Monday, October 15, 2012 4:18:30 PM UTC-5, Dean Michael Berris wrote:
>>
>> On Sun, Oct 14, 2012 at 2:58 PM, Matt Clark <ryu.t...@gmail.com> wrote:
>> > Oooook, so I have some good new and some bad news.
>> >
>> > Good news:
>> > -I have the lib compiling...
>> >
>>
>> Cool!
>>
>> > Bad news:
>> > -I had to install a VM of linux to do so :-/
>>
>> Hmmm... that's not good.
>>
>> > -2 tests fail in the make test
>> >
>>
>> I don't see these on the Mac though. Is this ToT Clang+libc++?
>
>
> TBH I don't know that it has anything to do with that. I think it has more
> to do with where you live/how your network is set up(at least for one of the
> failed tests), but I could be wrong.
>

Right. I thought you were having issues with the build/link -- they
seem to have been fixed now?

> Here's why I think that:
>
> Here is the jist of the log from the test:
>
> 15/21 Testing: cpp-netlib-http-client_get_test
> Running 2 test cases...
> /cpp-netlib/libs/network/test/http/client_get_test.cpp:29: error in
> "http_client_get_test": check status_ == 302u failed [200 != 302]
> /cpp-netlib/libs/network/test/http/client_get_test.cpp:30: error in
> "http_client_get_test": check status_message_ == std::string("Found") failed
> [OK != Found]
> /cpp-netlib/libs/network/test/http/client_get_test.cpp:50: error in
> "https_client_get_test": check status_ == 302u failed [200 != 302]
> /cpp-netlib/libs/network/test/http/client_get_test.cpp:51: error in
> "https_client_get_test": check status_message_ == std::string("Found")
> failed [OK != Found]
>
> Now in the code for that test 302u and std::string("Found") are both hard
> coded. So I'm getting a straight up OK 200 status from the connection to
> "http://www.google.com" which is a good thing^^ It seems you are used to
> getting a redirecting 302 Found with your setup, most likely because you
> aren't in the US and are therefore getting redirected to the regional
> version of google^^
>
> I could be wrong, but I think that this may be better stated as
> BOOST_CHECK(status_==302u||status_==200) and BOOST_CHECK(status_message_ ==
> std::string("Found")||status_message_ == std::string("OK")) .
>
> What do you think?
>

Sounds good to me.

> I'll be happy to test it do a pull request if you would like^^
>

Yes, please, that would be most appreciated.

> Cheers,
> -Matt
>
> PS Just got the news today that if we haven't had our child by Tuesday the
> 23rd the doctors will be inducing the birth^^
>

Congratulations in advance!

[Matt Clark]

Pull request submitted. The other tests that are failing on that build are do to the boost version I'm using, so I'll fix that and see if I can't start using the lib and making some progress!^^

[Dean Michael Berris]

On Tue, Oct 16, 2012 at 11:12 PM, Matt Clark <ryu.t...@gmail.com> wrote:
> Pull request submitted.

And merged! :D

> The other tests that are failing on that build are
> do to the boost version I'm using, so I'll fix that and see if I can't start
> using the lib and making some progress!^^
>

Cool, thanks Matt! :)