You can apply the visitor to the variant and receive the instance of the stored type. With Kind Regards, Ovanes On Tue, May 6, 2008 at 3:05 PM, Igor R. wrote:

Hi all,

How can I get the first "bounded type" of a variant? I see there's a public typedef there:

typedef typename mpl::transform< recursive_enabled_types , unwrap_recursivempl::_1 >::type types;

but I don't realize how I can "extract" the type from here. I need something like this:

MyVariantType v; MyVariantType::FirstBoundedType *value = getMyVariantType::FirstBoundedType(&v);

Thank you!

------------------------------ Invite your mail contacts to join your friends list with Windows Live Spaces. It's easy! Try it!http://spaces.live.com/spacesapi.aspx?wx_action=create&wx_url=/friends.aspx&mkt=en-us

_______________________________________________ Boost-users mailing list Boost-users@lists.boost.org http://lists.boost.org/mailman/listinfo.cgi/boost-users

AMDG Igor R. wrote:

Hi Ovanes,

Yes, I know, but that's not what need.

#include mpl::frontMyVariant::types::type is the type of the first element. In Christ, Steven Watanabe

First I was going to suggest you to use mpl::front to retrieve the first element of the type sequence, but the new variant doc does not state that the first element of the type sequence will be created in the variant's memory. I remember to read that it was once the case. You can try this: (But I am not sure if it works or is portable) typedef typename mpl::transform< recursive_enabled_types , unwrap_recursivempl::_1 >::type types; typedef boost::mpl::front<types>::type first_type; MyVariantType v; first_type *value = get(&v); Regards, Ovanes On Tue, May 6, 2008 at 5:33 PM, Igor R. wrote:

Hi Ovanes,

Yes, I know, but that's not what need.

------------------------------ Date: Tue, 6 May 2008 17:25:08 +0200 From: om_boost@keywallet.com To: boost-users@lists.boost.org Subject: Re: [Boost-users] [variant] how to get 1st bounded type

You can apply the visitor to the variant and receive the instance of the stored type.

With Kind Regards, Ovanes

2008/5/6 Igor R. :

...

First I was going to suggest you to use mpl::front to retrieve the first element of the type sequence, but the new variant doc does not state that the first element of the type sequence will be created in the variant's memory. I remember to read that it was once the case.

At least, the document that appears at the site states so: [...]

http://www.boost.org/doc/libs/1_35_0/doc/html/variant/design.html#variant.de...

At least what I read here is following:

The Guarantee All instances v of type varianthttp://www.boost.org/doc/libs/1_35_0/doc/html/boost/variant.html guarantee that v has constructed content of !!!one!!! of the types !!!T*i!!!*, even if an operation on v has previously failed. For me it means that it can be T2 or TN as well. Regards, Ovanes

On Tue, May 6, 2008 at 6:31 PM, Igor R. wrote:

...

At least what I read here is following: All instances v of type varianthttp://www.boost.org/doc/libs/1_35_0/doc/html/boost/variant.html guarantee that v has constructed content of !!!one!!! of the types !!!T*i!!!*, even if an operation on v has previously failed. For me it means that it can be T2 or TN as well.

Right, but the 1st link says: "*boost::variant*http://www.boost.org/doc/libs/1_35_0/doc/html/boost/variant.html< int, std::string > v; By default, a variant default-constructs its first bounded type, so vinitially contains int(0). If this is not desired, or if the first bounded type is not default-constructible, a variant can be constructed directly from any value convertible to one of its bounded types. "

Igor, I assume the following:

If the first type throws an exception while creation the second is created and so forth and the exception is probably not reported to you. Your pointer can be NULL in your previous example. Why don't you like visitor to dispatch the type stored in the variant? Regards, Ovanes

Hi Ovanes,

Igor, I assume the following:> If the first type throws an exception while creation the second is created and so forth and the exception is probably not reported to you.

No, as Steven aready wrote, that's not what happens.

Why don't you like visitor to dispatch the type stored in the variant?

I like it very much, and I use variant because of the power of the static visiting mechanism. The reason I wanted to test if the variant contains its 1st type is the following: my variant is made of a list of shared-pointers to various types; there's a routine that initializes it with one of the types depending on some conditions, then it applies some visitor on the resulting variant and stores it then in a container - however, there's no "default" behavior, so if neither condition was met, the variant remains "uninitialized" - i.e. it contains 1st type "zero" ptr. I want neither apply the visitor on "empty" smartpointer nor store it in the container, so I have to know whether it was "really" created or not. Two simple ways I can think about are to test the resulting variant or to introduce some flag - of course, both ways are far from ideal... _________________________________________________________________ Connect to the next generation of MSN Messenger  http://imagine-msn.com/messenger/launch80/default.aspx?locale=en-us&source=wlmailtagline

Oop! sorry, I did not intend to send that yet -- I was toying with the problem discussed in this thread but my finger slipped.. :( ________________________________ From: John Femiani Sent: Wednesday, May 07, 2008 2:53 AM To: 'boost-users@lists.boost.org' Subject: RE: [Boost-users] [variant] how to get 1st bounded type > Why don't you like visitor to dispatch the type stored in the variant? I like it very much, and I use variant because of the power of the static visiting mechanism. The reason I wanted to test if the variant contains its 1st type is the following: my variant is made of a list of shared-pointers to various types; there's a routine that initializes it with one of the types depending on some conditions, then it applies some visitor on the resulting variant and stores it then in a container - however, there's no "default" behavior, so if neither condition was met, the variant remains "uninitialized" - i.e. it contains 1st type "zero" ptr. I want neither apply the visitor on "empty" smartpointer nor store it in the container, so I have to know whether it was "really" created or not. Two simple ways I can think about are to test the resulting variant or to introduce some flag - of course, both ways are far from ideal... struct my_visitor { typedef bool return_type; vector & vec_; my_visitor (vector & vec) : vec_(vec) {} my_visitor (my_visitor & other) : vec_(other.vec_) {} bool opeator()(...) {return false;} bool operator()(type1 value) { vec_.push_back(variant_type(value)); return true; } bool operator()(type2 value) { vec_.push_back(variant_type(value)); return true; } } vector my_vec :boost::apply_visitor(my_visitor(my_vec), my_variant)

Hmm... how could I miss such a straightforward solution?? :) Thank you!> Date: Wed, 7 May 2008 22:18:11 -0400> From: nat@lindenlab.com> To: boost-users@lists.boost.org> Subject: Re: [Boost-users] [variant] how to get 1st bounded type> > Igor R. wrote:> > > The reason I wanted to test if the variant contains its 1st> > type is the following: my variant is made of a list> > of shared-pointers to various types; there's a routine that> > initializes it with one of the types depending on some> > conditions, then it applies some visitor on the resulting> > variant and stores it then in a container - however, there's> > no "default" behavior, so if neither condition was met, the> > variant remains "uninitialized" - i.e. it contains 1st type> > "zero" ptr. I want neither apply the visitor on "empty"> > smartpointer nor store it in the container, so I have to> > know whether it was "really" created or not. > > Um, wrap the variant in boost::optional?> _______________________________________________> Boost-users mailing list> Boost-users@lists.boost.org> http://lists.boost.org/mailman/listinfo.cgi/boost-users _________________________________________________________________ Discover the new Windows Vista http://search.msn.com/results.aspx?q=windows+vista&mkt=en-US&form=QBRE

Max wrote:

...

Hmm... how could I miss such a straightforward solution?? :)

how is this solution straightforward? Please elaborate.

The OP stated that sometimes he needs his type to express "none of the above." boost::optional is specifically intended to distinguish the case in which you have no meaningful value -- rather like SQL's NULL.

AMDG Ovanes Markarian wrote:

At least what I read here is following:

The Guarantee

All instances |v| of type |variant http://www.boost.org/doc/libs/1_35_0/doc/html/boost/variant.html| guarantee that |v| has constructed content of !!!one!!! of the types |!!!T/i!!!/|, even if an operation on |v| has previously failed.

For me it means that it can be T2 or TN as well.

If, variant could only hold the first type there would be no reason to use it. The whole point of variant is that it can hold any of the types in the list. In Christ, Steven Watanabe