[Range] and [MemberFunction]

older
Newbie Question - Errors if i want...

Robert Jones

13 Mar 2012 13 Mar '12

2:29 p.m.

In the documentation of Boost.MemberFunction it suggests that a useful purpose for it would be to implement overloads of std algorithms, and indeed other similar algorithms such that the final callable parameter could be a simple pointer to member. struct X { bool method( ); }; std::vector<X> xs; for_each( xs.begin( ), xs.end( ), & X::method ); or from Boost.Range boost::for_each( xs, &X::method ); Is such a facility implemented anywhere in Boost? Where I'm really going with this, is that I'd like to see versions of Boost.Range adaptors which support the same facility, so I can say, for example for_each( xs | filtered( & X::method ), doSomething ); rather than, as at present, for_each ( xs | filtered( bind( & X::method, _1 ) ), doSomething ); and similarly for transformed. Thanks, - Rob.

Attachments:

attachment.html (text/html — 1.6 KB)

Show replies by date

Ovanes Markarian

13 Mar 13 Mar

3:05 p.m.

On Tue, Mar 13, 2012 at 3:29 PM, Robert Jones wrote:

...

In the documentation of Boost.MemberFunction it suggests that a useful purpose for it would be to implement overloads of std algorithms, and indeed other similar algorithms such that the final callable parameter could be a simple pointer to member.

struct X { bool method( ); };

std::vector<X> xs;

for_each( xs.begin( ), xs.end( ), & X::method );

or from Boost.Range

boost::for_each( xs, &X::method );

Is such a facility implemented anywhere in Boost?

Where I'm really going with this, is that I'd like to see versions of Boost.Range adaptors which support the same facility, so I can say, for example

for_each( xs | filtered( & X::method ), doSomething );

rather than, as at present,

for_each ( xs | filtered( bind( & X::method, _1 ) ), doSomething );

and similarly for transformed.

Thanks,

- Rob.

What about Boost.Foreach??? http://www.boost.org/doc/libs/1_49_0/doc/html/foreach.html Regards, ovanes

Robert Jones

3:16 p.m.

On Tue, Mar 13, 2012 at 3:05 PM, Ovanes Markarian wrote:

...

What about Boost.Foreach??? http://www.boost.org/doc/libs/1_49_0/doc/html/foreach.html

I can see how that addresses the 'for_each' examples, though it sidesteps a bit, but not the filtered and transformed issues, which is the real deal for me. Thx, R.

Ovanes Markarian

3:26 p.m.

On Tue, Mar 13, 2012 at 4:16 PM, Robert Jones wrote:

...

On Tue, Mar 13, 2012 at 3:05 PM, Ovanes Markarian wrote:

...
What about Boost.Foreach??? http://www.boost.org/doc/libs/1_49_0/doc/html/foreach.html

I can see how that addresses the 'for_each' examples, though it sidesteps a bit, but not the filtered and transformed issues, which is the real deal for me.

Thx, R.

Boost offers filter and transform iterators, which can be put into the range ;) http://www.boost.org/doc/libs/1_49_0/libs/iterator/doc/index.html Best Regards, Ovanes

Robert Jones

3:37 p.m.

On Tue, Mar 13, 2012 at 3:26 PM, Ovanes Markarian wrote:

...

On Tue, Mar 13, 2012 at 4:16 PM, Robert Jones wrote:

...
On Tue, Mar 13, 2012 at 3:05 PM, Ovanes Markarian
...
wrote:

...
What about Boost.Foreach??? http://www.boost.org/doc/libs/1_49_0/doc/html/foreach.html

I can see how that addresses the 'for_each' examples, though it sidesteps a bit, but not the filtered and transformed issues, which is the real deal for me.

Thx, R.

Boost offers filter and transform iterators, which can be put into the range ;)

http://www.boost.org/doc/libs/1_49_0/libs/iterator/doc/index.html

Indeed, however if I read it right filter_iterator requires a predicate, just as boost.adaptors.filtered does, and the construction of the predicate from a method requires the explicit use of bind, or mem_fn, whereas I'd hoped to see a facility offering an overload that accepts a pointer-to-member directly. Thx, R.

Philipp Moeller

3:36 p.m.

Robert Jones writes:

...

In the documentation of Boost.MemberFunction it suggests that a useful purpose for it would be to implement overloads of std algorithms, and indeed other similar algorithms such that the final callable parameter could be a simple pointer to member.

struct X { bool method( ); };

std::vector<X> xs;

for_each( xs.begin( ), xs.end( ), & X::method );

or from Boost.Range

boost::for_each( xs, &X::method );

Is such a facility implemented anywhere in Boost?

Where I'm really going with this, is that I'd like to see versions of Boost.Range adaptors which support the same facility, so I can say, for example

for_each( xs | filtered( & X::method ), doSomething );

rather than, as at present,

for_each ( xs | filtered( bind( & X::method, _1 ) ), doSomething );

and similarly for transformed.

Thanks,

- Rob.

I don't know Boost.MemberFunction and couldn't find any documentation for it, could you provide a link? But from what I can gather in your message, boost::mem_fn seems to be exactly this: for_each( xs.begin( ), xs.end( ), boost::mem_fn(& X::method) ); Granted, it is not exactly what you want, but gets very close. HTH, Philipp Moeller

Robert Jones

3:43 p.m.

On Tue, Mar 13, 2012 at 3:36 PM, Philipp Moeller < philipp.moeller@geometryfactory.com> wrote:

...

I don't know Boost.MemberFunction and couldn't find any documentation for it, could you provide a link?

But from what I can gather in your message, boost::mem_fn seems to be exactly this:

for_each( xs.begin( ), xs.end( ), boost::mem_fn(& X::method) );

Granted, it is not exactly what you want, but gets very close.

You're there already mem_fn is called 'Member Function' in the front page list. It is exactly the boost::mem_fn syntactic noise I was seeking to avoid. - R.

Philipp Moeller

6:07 p.m.

Robert Jones writes:

...

On Tue, Mar 13, 2012 at 3:36 PM, Philipp Moeller < philipp.moeller@geometryfactory.com> wrote:

...
I don't know Boost.MemberFunction and couldn't find any documentation for it, could you provide a link?

But from what I can gather in your message, boost::mem_fn seems to be exactly this:

for_each( xs.begin( ), xs.end( ), boost::mem_fn(& X::method) );

Granted, it is not exactly what you want, but gets very close.

You're there already mem_fn is called 'Member Function' in the front page list.

It is exactly the boost::mem_fn syntactic noise I was seeking to avoid.

- R.

MaximYanchenko

14 Mar 14 Mar

2:16 p.m.

Robert Jones writes:

...

Where I'm really going with this, is that I'd like to see versions of Boost.Range adaptors which support the same facility

Hi Rob, You can find this patch useful: http://boost.2283326.n4.nabble.com/range-adaptors-and-member-functions-point... I hope it will be accepted into Boost.Range some time. Thanks, Maxim

Robert Jones

16 Mar 16 Mar

8:16 a.m.

On Wed, Mar 14, 2012 at 2:16 PM, MaximYanchenko wrote:

...

Robert Jones writes:

...
Where I'm really going with this, is that I'd like to see versions of Boost.Range adaptors which support the same facility

Hi Rob, You can find this patch useful:

http://boost.2283326.n4.nabble.com/range-adaptors-and-member-functions-point...

I hope it will be accepted into Boost.Range some time.

Thx Maxim, mainly because I was beginning to think I was talking gibberish!

- Rob.

Robert Jones

9 a.m.

On Wed, Mar 14, 2012 at 2:16 PM, MaximYanchenko wrote:

...

Robert Jones writes:

...
Where I'm really going with this, is that I'd like to see versions of Boost.Range adaptors which support the same facility

Hi Rob, You can find this patch useful:

http://boost.2283326.n4.nabble.com/range-adaptors-and-member-functions-point...

I hope it will be accepted into Boost.Range some time.

Hi Maxim - Now that I've digested your patch and the thread you linked to, a question... The patch you've submitted will apply to transformed *and* filtered, possibly others too? Thx, R.

4622

Age (days ago)

4625

Last active (days ago)

List overview

Download

10 comments

4 participants

participants (4)

MaximYanchenko
Ovanes Markarian
Philipp Moeller
Robert Jones