Mirror of the gdb-patches mailing list
 help / color / mirror / Atom feed
* [patch 2/3] Extend C++ import to support renaming even declarations
@ 2010-04-30 18:16 Jan Kratochvil
  2010-04-30 18:54 ` Tom Tromey
  0 siblings, 1 reply; 5+ messages in thread
From: Jan Kratochvil @ 2010-04-30 18:16 UTC (permalink / raw)
  To: gdb-patches; +Cc: Sami Wagiaalla

Hi,

the current C++ using-directive supports renaming of namespaces but it does
not support renaming of specific imported declarations.  This is fine as C++
does not support the latter.

The Fortran patch 3/3 uses the existing FSF GDB C++ infrastructure, therefore
it had to extend it.

C++:     using namespace A;
Fortran: use A
	 - namespace import, therefore not a "declaration" import
	 - no renaming happens

C++:     using A::x;
Fortran: use A, only: x
	 - "declaration" import of the specific variable `x'.
	 - no renaming happens

namespace B = A;
Fortran has no way to address non-local namespace/module.
	 - namespace import, therefore not a "declaration" import
	 - renaming "A" to "B"

C++ cannot, it would be something like: using y = A::x;
Fortran: use A, only y => x
	 - "declaration" import of the specific variable `x'.
	 - renaming "x" to "y"

Therefore this patch implements the last case.  It does not affect existing
C++ functionality in the other 3 cases.

No regressions on {x86_64,x86_64-m32,i686}-fedora12-linux-gnu for the whole
patch series.


Thanks,
Jan


2010-04-30  Jan Kratochvil  <jan.kratochvil@redhat.com>

	* cp-namespace.c (cp_lookup_symbol_imports): Support ALIAS for the
	CURRENT->DECLARATION case.

--- a/gdb/cp-namespace.c
+++ b/gdb/cp-namespace.c
@@ -355,12 +355,14 @@ cp_lookup_symbol_imports (const char *scope,
 	searched_cleanup = make_cleanup (reset_directive_searched, current);
 
 	/* If there is an import of a single declaration, compare the imported
-	   declaration with the sought out name.  If there is a match pass
-	   current->import_src as NAMESPACE to direct the search towards the
-	   imported namespace.  */
-	if (current->declaration && strcmp (name, current->declaration) == 0)
+	   declaration (after optional renaming by its alias) with the sought
+	   out name.  If there is a match pass current->import_src as NAMESPACE
+	   to direct the search towards the imported namespace.  */
+	if (current->declaration
+	    && strcmp (name, current->alias ? current->alias
+					    : current->declaration) == 0)
 	  sym = cp_lookup_symbol_in_namespace (current->import_src,
-	                                       name,
+	                                       current->declaration,
 	                                       block,
 	                                       domain);
 


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [patch 2/3] Extend C++ import to support renaming even declarations
  2010-04-30 18:16 [patch 2/3] Extend C++ import to support renaming even declarations Jan Kratochvil
@ 2010-04-30 18:54 ` Tom Tromey
  2010-04-30 21:34   ` Jan Kratochvil
  0 siblings, 1 reply; 5+ messages in thread
From: Tom Tromey @ 2010-04-30 18:54 UTC (permalink / raw)
  To: Jan Kratochvil; +Cc: gdb-patches, Sami Wagiaalla

>>>>> "Jan" == Jan Kratochvil <jan.kratochvil@redhat.com> writes:

Jan> Therefore this patch implements the last case.  It does not affect existing
Jan> C++ functionality in the other 3 cases.

Jan> 2010-04-30  Jan Kratochvil  <jan.kratochvil@redhat.com>

Jan> 	* cp-namespace.c (cp_lookup_symbol_imports): Support ALIAS for the
Jan> CURRENT-> DECLARATION case.

Please also update the documentation for struct using_direct to mention
this case.  Right now the comment there indicates that alias==NULL in
all other cases.

This patch is ok with that change.

I wonder why the fields of using_direct are not const char *.

Tom


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [patch 2/3] Extend C++ import to support renaming even declarations
  2010-04-30 18:54 ` Tom Tromey
@ 2010-04-30 21:34   ` Jan Kratochvil
  2010-05-03 18:02     ` Tom Tromey
  0 siblings, 1 reply; 5+ messages in thread
From: Jan Kratochvil @ 2010-04-30 21:34 UTC (permalink / raw)
  To: Tom Tromey; +Cc: gdb-patches, Sami Wagiaalla

On Fri, 30 Apr 2010 20:54:12 +0200, Tom Tromey wrote:
> Please also update the documentation for struct using_direct to mention
> this case.  Right now the comment there indicates that alias==NULL in
> all other cases.

OK this way?

(I understand the Fortran lines in the comment should be checked-in only with
the patch 3/3 implementing them.)


Thanks,
Jan


2010-04-30  Jan Kratochvil  <jan.kratochvil@redhat.com>

	* cp-namespace.c (cp_lookup_symbol_imports): Support ALIAS for the
	CURRENT->DECLARATION case.
	* cp-support.h (struct using_direct): Provide extended comment.

--- a/gdb/cp-namespace.c
+++ b/gdb/cp-namespace.c
@@ -355,12 +355,14 @@ cp_lookup_symbol_imports (const char *scope,
 	searched_cleanup = make_cleanup (reset_directive_searched, current);
 
 	/* If there is an import of a single declaration, compare the imported
-	   declaration with the sought out name.  If there is a match pass
-	   current->import_src as NAMESPACE to direct the search towards the
-	   imported namespace.  */
-	if (current->declaration && strcmp (name, current->declaration) == 0)
+	   declaration (after optional renaming by its alias) with the sought
+	   out name.  If there is a match pass current->import_src as NAMESPACE
+	   to direct the search towards the imported namespace.  */
+	if (current->declaration
+	    && strcmp (name, current->alias ? current->alias
+					    : current->declaration) == 0)
 	  sym = cp_lookup_symbol_in_namespace (current->import_src,
-	                                       name,
+	                                       current->declaration,
 	                                       block,
 	                                       domain);
 
--- a/gdb/cp-support.h
+++ b/gdb/cp-support.h
@@ -37,19 +37,44 @@ struct type;
 struct demangle_component;
 
 /* This struct is designed to store data from using directives.  It
-   says that names from namespace IMPORT_SRC should be visible within
-   namespace IMPORT_DEST.  These form a linked list; NEXT is the next element
-   of the list.  If the imported namespace has been aliased, ALIAS is set to a
-   string representing the alias.  Otherwise, ALIAS is NULL.
-   Eg:
-       namespace C = A::B;
-   ALIAS = "C"
-   DECLARATION is the name of the imported declaration, if this import
-   statement represents one.
-   Eg:
-       using A::x;
-   Where x is variable in namespace A.  DECLARATION is set to x.
-*/
+   says that names from namespace IMPORT_SRC should be visible within namespace
+   IMPORT_DEST.  These form a linked list; NEXT is the next element of the
+   list.  If the imported namespace or declaration has been aliased within the
+   IMPORT_DEST namespace, ALIAS is set to a string representing the alias.
+   Otherwise, ALIAS is NULL.  DECLARATION is the name of the imported
+   declaration, if this import statement represents one.  Otherwise DECLARATION
+   is NULL and this import statement represents a namespace.
+
+   C++:      using namespace A;
+   Fortran:  use A
+   import_src = "A"
+   import_dest = local scope of the import statement even such as ""
+   alias = NULL
+   declaration = NULL
+
+   C++:      using A::x;
+   Fortran:  use A, only: x
+   import_src = "A"
+   import_dest = local scope of the import statement even such as ""
+   alias = NULL
+   declaration = "x"
+   The declaration will get imported as import_dest::x.
+
+   C++:      namespace LOCALNS = A;
+   Fortran has no way to address non-local namespace/module.
+   import_src = "A"
+   import_dest = local scope of the import statement even such as ""
+   alias = "LOCALNS"
+   declaration = NULL
+   The namespace will get imported as the import_dest::LOCALNS namespace.
+
+   C++ cannot express it, it would be something like:  using localname = A::x;
+   Fortran:  use A, only localname => x
+   import_src = "A"
+   import_dest = local scope of the import statement even such as ""
+   alias = "localname"
+   declaration = "x"
+   The declaration will get imported as localname or `import_dest`localname.  */
 
 struct using_direct
 {


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [patch 2/3] Extend C++ import to support renaming even declarations
  2010-04-30 21:34   ` Jan Kratochvil
@ 2010-05-03 18:02     ` Tom Tromey
  2010-05-03 20:11       ` Jan Kratochvil
  0 siblings, 1 reply; 5+ messages in thread
From: Tom Tromey @ 2010-05-03 18:02 UTC (permalink / raw)
  To: Jan Kratochvil; +Cc: gdb-patches, Sami Wagiaalla

>>>>> "Jan" == Jan Kratochvil <jan.kratochvil@redhat.com> writes:

Jan> OK this way?

Yes, thanks.

Tom


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [patch 2/3] Extend C++ import to support renaming even declarations
  2010-05-03 18:02     ` Tom Tromey
@ 2010-05-03 20:11       ` Jan Kratochvil
  0 siblings, 0 replies; 5+ messages in thread
From: Jan Kratochvil @ 2010-05-03 20:11 UTC (permalink / raw)
  To: Tom Tromey; +Cc: gdb-patches, Sami Wagiaalla

On Mon, 03 May 2010 20:02:30 +0200, Tom Tromey wrote:
> >>>>> "Jan" == Jan Kratochvil <jan.kratochvil@redhat.com> writes:
> 
> Jan> OK this way?
> 
> Yes, thanks.

Checked-in:
	http://sourceware.org/ml/gdb-cvs/2010-05/msg00021.html


Thanks,
Jan


^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2010-05-03 20:11 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2010-04-30 18:16 [patch 2/3] Extend C++ import to support renaming even declarations Jan Kratochvil
2010-04-30 18:54 ` Tom Tromey
2010-04-30 21:34   ` Jan Kratochvil
2010-05-03 18:02     ` Tom Tromey
2010-05-03 20:11       ` Jan Kratochvil

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox