|
From: Gustavo C. <gjc...@gm...> - 2008-06-29 16:03:20
|
I have a function[1] that extracts properties from a type, using
pygccxml.declarations.type_traits
For a simple type, it does nothing, and the type name is unchanged. All is
well here:
***** type traits for '::uint64_t'
*** > return ('uint64_t', False, 0, False)
In the following case, I call type_traits.remove_const and
type_traits.remove_reference, and the resulting type then appears different:
***** type traits for '::uint64_t const &'
*** > return ('long unsigned int', True, 0, True)
Yes, I know that in this system "long unsigned int" is the same as
uint64_t. But I would rather work with uint64_t, due to LLP/LP platform
differences.
Any hints on how to work around this?
[1]
http://bazaar.launchpad.net/~gjc/pybindgen/trunk/annotate/gjc%40gnome.org-20080629154751-9lgrks1m411ssn8f?file_id=gccxmlparser.py-20071117182614-93x4j7mcuw0eq1q6-1#L135
--
Gustavo J. A. M. Carneiro
INESC Porto, Telecommunications and Multimedia Unit
"The universe is always one step beyond logic." -- Frank Herbert
|
|
From: Roman Y. <rom...@gm...> - 2008-06-29 17:21:37
|
On Sun, Jun 29, 2008 at 7:03 PM, Gustavo Carneiro <gjc...@gm...> wrote:
> I have a function[1] that extracts properties from a type, using
> pygccxml.declarations.type_traits
>
> For a simple type, it does nothing, and the type name is unchanged. All is
> well here:
>
> ***** type traits for '::uint64_t'
> *** > return ('uint64_t', False, 0, False)
>
>
> In the following case, I call type_traits.remove_const and
> type_traits.remove_reference, and the resulting type then appears different:
>
> ***** type traits for '::uint64_t const &'
> *** > return ('long unsigned int', True, 0, True)
>
> Yes, I know that in this system "long unsigned int" is the same as
> uint64_t. But I would rather work with uint64_t, due to LLP/LP platform
> differences.
>
> Any hints on how to work around this?
Yes. You can improve the function. The first thing the current
implementation does - it removes "typedef"'s, it simplifies a lot.
Almost all functions in type_traits implemented this way.
I am a little bit busy these days, if you can wait, I will take a look
on it. If not the patch is welcome.
--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/
|
|
From: Gustavo C. <gjc...@gm...> - 2008-06-29 20:56:57
|
2008/6/29 Roman Yakovenko <rom...@gm...>:
> On Sun, Jun 29, 2008 at 7:03 PM, Gustavo Carneiro <gjc...@gm...>
> wrote:
> > I have a function[1] that extracts properties from a type, using
> > pygccxml.declarations.type_traits
> >
> > For a simple type, it does nothing, and the type name is unchanged. All
> is
> > well here:
> >
> > ***** type traits for '::uint64_t'
> > *** > return ('uint64_t', False, 0, False)
> >
> >
> > In the following case, I call type_traits.remove_const and
> > type_traits.remove_reference, and the resulting type then appears
> different:
> >
> > ***** type traits for '::uint64_t const &'
> > *** > return ('long unsigned int', True, 0, True)
> >
> > Yes, I know that in this system "long unsigned int" is the same as
> > uint64_t. But I would rather work with uint64_t, due to LLP/LP platform
> > differences.
> >
> > Any hints on how to work around this?
>
> Yes. You can improve the function. The first thing the current
> implementation does - it removes "typedef"'s, it simplifies a lot.
> Almost all functions in type_traits implemented this way.
>
> I am a little bit busy these days, if you can wait, I will take a look
> on it. If not the patch is welcome.
What I did was the obvious thing. Removed the remove_alias call, and it now
works perfectly for me :-)
I honestly don't understand why remove_alias is needed here :-/
Index: pygccxml/declarations/type_traits.py
===================================================================
--- pygccxml/declarations/type_traits.py (revision 1362)
+++ pygccxml/declarations/type_traits.py (working copy)
@@ -171,7 +171,8 @@
If type is not pointer type, it will be returned as is.
"""
- nake_type = remove_alias( type )
+ #nake_type = remove_alias( type )
+ nake_type = type
if not is_pointer( nake_type ):
return type
elif isinstance( nake_type, cpptypes.volatile_t ) and isinstance(
nake_type.base, cpptypes.pointer_t ):
@@ -219,7 +220,8 @@
If type is not reference type, it will be returned as is.
"""
- nake_type = remove_alias( type )
+ #nake_type = remove_alias( type )
+ nake_type = type
if not is_reference( nake_type ):
return type
else:
@@ -236,7 +238,8 @@
If type is not const type, it will be returned as is
"""
- nake_type = remove_alias( type )
+ #nake_type = remove_alias( type )
+ nake_type = type
if not is_const( nake_type ):
return type
else:
>
>
> --
> Roman Yakovenko
> C++ Python language binding
> http://www.language-binding.net/
>
--
Gustavo J. A. M. Carneiro
INESC Porto, Telecommunications and Multimedia Unit
"The universe is always one step beyond logic." -- Frank Herbert
|
|
From: Roman Y. <rom...@gm...> - 2008-06-30 05:15:59
|
On Sun, Jun 29, 2008 at 11:57 PM, Gustavo Carneiro <gjc...@gm...> wrote: > What I did was the obvious thing. Removed the remove_alias call, and it now > works perfectly for me :-) > > I honestly don't understand why remove_alias is needed here :-/ typedef const XYZ CXYZ; remove_const( CXYZ ) will not work in such case and obviously you have to take into account few levels of "typedef"'s > > Index: pygccxml/declarations/type_traits.py > =================================================================== > --- pygccxml/declarations/type_traits.py (revision 1362) > +++ pygccxml/declarations/type_traits.py (working copy) > @@ -171,7 +171,8 @@ > > If type is not pointer type, it will be returned as is. > """ > - nake_type = remove_alias( type ) > + #nake_type = remove_alias( type ) > + nake_type = type > if not is_pointer( nake_type ): > return type > elif isinstance( nake_type, cpptypes.volatile_t ) and isinstance( > nake_type.base, cpptypes.pointer_t ): > @@ -219,7 +220,8 @@ > > If type is not reference type, it will be returned as is. > """ > - nake_type = remove_alias( type ) > + #nake_type = remove_alias( type ) > + nake_type = type > if not is_reference( nake_type ): > return type > else: > @@ -236,7 +238,8 @@ > If type is not const type, it will be returned as is > """ > > - nake_type = remove_alias( type ) > + #nake_type = remove_alias( type ) > + nake_type = type > if not is_const( nake_type ): > return type > else: > > >> >> -- >> Roman Yakovenko >> C++ Python language binding >> http://www.language-binding.net/ > > > > -- > Gustavo J. A. M. Carneiro > INESC Porto, Telecommunications and Multimedia Unit > "The universe is always one step beyond logic." -- Frank Herbert -- Roman Yakovenko C++ Python language binding http://www.language-binding.net/ |
|
From: Gustavo C. <gjc...@gm...> - 2008-06-30 12:11:43
|
2008/6/30 Roman Yakovenko <rom...@gm...>: > On Sun, Jun 29, 2008 at 11:57 PM, Gustavo Carneiro <gjc...@gm...> > wrote: > > What I did was the obvious thing. Removed the remove_alias call, and it > now > > works perfectly for me :-) > > > > I honestly don't understand why remove_alias is needed here :-/ > > typedef const XYZ CXYZ; > > remove_const( CXYZ ) > > > will not work in such case and obviously you have to take into account > few levels of "typedef"'s I don't actually _need_ to remove consts from typedefs, only from the outer type. Anyway, I copy-pasted those functions into pybindgen and patched them to do what I need, so all is fine now. Thanks anyway. > > > > > > Index: pygccxml/declarations/type_traits.py > > =================================================================== > > --- pygccxml/declarations/type_traits.py (revision 1362) > > +++ pygccxml/declarations/type_traits.py (working copy) > > @@ -171,7 +171,8 @@ > > > > If type is not pointer type, it will be returned as is. > > """ > > - nake_type = remove_alias( type ) > > + #nake_type = remove_alias( type ) > > + nake_type = type > > if not is_pointer( nake_type ): > > return type > > elif isinstance( nake_type, cpptypes.volatile_t ) and isinstance( > > nake_type.base, cpptypes.pointer_t ): > > @@ -219,7 +220,8 @@ > > > > If type is not reference type, it will be returned as is. > > """ > > - nake_type = remove_alias( type ) > > + #nake_type = remove_alias( type ) > > + nake_type = type > > if not is_reference( nake_type ): > > return type > > else: > > @@ -236,7 +238,8 @@ > > If type is not const type, it will be returned as is > > """ > > > > - nake_type = remove_alias( type ) > > + #nake_type = remove_alias( type ) > > + nake_type = type > > if not is_const( nake_type ): > > return type > > else: > > > > > >> > >> -- > >> Roman Yakovenko > >> C++ Python language binding > >> http://www.language-binding.net/ > > > > > > > > -- > > Gustavo J. A. M. Carneiro > > INESC Porto, Telecommunications and Multimedia Unit > > "The universe is always one step beyond logic." -- Frank Herbert > > > > -- > Roman Yakovenko > C++ Python language binding > http://www.language-binding.net/ > -- Gustavo J. A. M. Carneiro INESC Porto, Telecommunications and Multimedia Unit "The universe is always one step beyond logic." -- Frank Herbert |
|
From: Roman Y. <rom...@gm...> - 2008-06-30 12:26:53
|
On Mon, Jun 30, 2008 at 3:11 PM, Gustavo Carneiro <gjc...@gm...> wrote:
> I don't actually _need_ to remove consts from typedefs, only from the outer
> type.
I am not sure you are right. Consider next use case:
struct item_t{};
typedef const item_t c_item_t;
struct value_t{
c_item_t v;
};
Obviously, in this case you have to find out whether the "v" is const
or not, and generate different code.
I guess you are talking about some specific use case, which is not this.
> Anyway, I copy-pasted those functions into pybindgen and patched them to do
> what I need, so all is fine now.
:-(. I will see what I can do.
--
Roman Yakovenko
C++ Python language binding
http://www.language-binding.net/
|
|
From: Gustavo C. <gjc...@gm...> - 2008-06-30 12:33:51
|
2008/6/30 Roman Yakovenko <rom...@gm...>:
> On Mon, Jun 30, 2008 at 3:11 PM, Gustavo Carneiro <gjc...@gm...>
> wrote:
> > I don't actually _need_ to remove consts from typedefs, only from the
> outer
> > type.
>
> I am not sure you are right. Consider next use case:
>
> struct item_t{};
>
> typedef const item_t c_item_t;
>
> struct value_t{
> c_item_t v;
> };
>
> Obviously, in this case you have to find out whether the "v" is const
> or not, and generate different code.
I see.
I think you're right, but that typedef is kind of evil anyway, and I would
rather deal with corner cases as they appear, particularly since some corner
cases often are theoretical and never manifest themselves in real life.
>
>
> I guess you are talking about some specific use case, which is not this.
>
> > Anyway, I copy-pasted those functions into pybindgen and patched them to
> do
> > what I need, so all is fine now.
>
> :-(. I will see what I can do.
Sorry. :P
If I get into trouble with my simplification, I'll help. Until then...
--
Gustavo J. A. M. Carneiro
INESC Porto, Telecommunications and Multimedia Unit
"The universe is always one step beyond logic." -- Frank Herbert
|
