Is using boost::variant an option? Then you could use something like
#include <boost/variant.hpp>
...
typedef boost::variant<uint8_t, bool> elem_type;
std::vector<elem_type> array1 = {uint8_t(0), uint8_t(1)};
std::vector<elem_type> array2 = {true, false};
std::vector<std::vector<elem_type>> array3 = {array1, array2};
This at least compiles, but I don't know if it will produce the results you
want from a QITS perspective.
Chris
On 28 March 2018 at 17:23, Kim van der Riet <kim.vdr...@redhat.com> wrote:
> This is part of the Qpid Interop Test Suite, a test written for complex
> type interoperability.
>
> AMQP allows arrays of arrays of differing types, I am wanting to test this
> using the different bindings against each other - this is what the QITS
> does.
>
> In the case of the C++ binding, the default encoding for an AMQP array
> occurs when a container (std::vector, std::deque, etc.) is of a c++ type
> rather than using proton::value, which would cause the encoding to be an
> AMQP list.
>
> Certainly I can see that the non-homogenous nature of the declaration is
> the issue, I just don't know how to get around it in a way that the Proton
> encoder will accept. I know there is a way to force a
> std::vector<proton::value> to be encoded as an array, but I don't know how
> to do that.
>
> Kim
>
>
> On 03/28/2018 11:49 AM, Chris Richardson wrote:
>
>> The short answer is "you can't", at least not along the lines you're
>> thinking. The std::vector type is a homogeneous container, see eg: this
>> for
>> more info: https://isocpp.org/wiki/faq/containers#heterogeneous-contain
>> ers
>>
>> The upshot is that the elements of the vectors must have a common (base)
>> type, which clearly uint8_t and bool do not. The usual way to get around
>> this is exactly as you've outlined - to use some sort of variant-style
>> container such as the proton::value or boost::variant etc. The
>> alternatives
>> (as mentioned in the above article) are almost too terrible to contemplate
>> and would involve lists of void pointers and risky reinterpret_casts
>> scattered through the code <shudder>.
>>
>> In some situations it can be useful to use a tuple type, then you could
>> have some construct like
>> std::tuple<std::vector<uint8_t>, std::vector<bool>>
>> but if your list isn't of a fixed size known at compile-time (or
>> specifically needs to be a vector) this probably isn't what you're after
>> either.
>>
>> Can you say more about what you're trying to do and why proton::value
>> isn't
>> sufficient?
>>
>> Chris
>>
>>
>>
>> On 28 March 2018 at 15:45, Kim van der Riet <kim.vdr...@redhat.com>
>> wrote:
>>
>> By default, using std::vector<c++-type> is encoded as an AMQP array,
>>> whereas std::vector<proton::value> is encoded as an AMQP list. So, to
>>> create an array of lists of differing types, for example:
>>>
>>> std::vector<proton::value> list1 = {uint8_t(0), uint8_t(1)};
>>> std::vector<proton::value> list2 = {true, false};
>>> std::vector<std::vector<proton::value> > array = {list1, list2};
>>>
>>> works ok. However, how does one define an array of arrays of differing
>>> types? For example:
>>>
>>> std::vector<uint8_t> array1 = {uint8_t(0), uint8_t(1)};
>>> std::vector<bool> array2 = {true, false};
>>> std::vector<std::vector<##> > array3 = {array1, array2};
>>>
>>> What is ##? Unless all the sub-arrays are of the same type, it is not
>>> possible to define a single type here. If I use proton::value (which
>>> would
>>> make it indistinguishable from the array of lists above), I get a compile
>>> error:
>>>
>>> could not convert '{array1, array2}' from '<brace-enclosed initializer
>>> list' to 'std::vector<std::vector<proton::value> >'
>>>
>>> What is the correct way to create an array of arrays of differing types?
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: users-unsubscr...@qpid.apache.org
>>> For additional commands, e-mail: users-h...@qpid.apache.org
>>>
>>>
>>>
>>
>>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscr...@qpid.apache.org
> For additional commands, e-mail: users-h...@qpid.apache.org
>
>
--
*Chris Richardson*, System Architect
c...@fourc.eu
*FourC AS, Vestre Rosten 81, Trekanten, NO-7075 Tiller, Norwaywww.fourc.eu
<http://www.fourc.eu/>*
*Follow us on LinkedIn <http://bit.ly/fourcli>, Facebook
<http://bit.ly/fourcfb>, Google+ <http://bit.ly/fourcgp> and Twitter
<http://bit.ly/fourctw>!*
