Differences between revisions 26 and 27
Revision 26 as of 2007-06-22 17:14:07
Size: 3762
Editor: TobiasBurnus
Comment: Link to Chart of Fortran 2003 Features supported by GNU Fortran
Revision 27 as of 2007-06-22 17:16:31
Size: 3808
Editor: TobiasBurnus
Comment: Link to SoC
Deletions are marked like this. Additions are marked like this.
Line 33: Line 33:
 * '''Procedure pointers'''  * '''Procedure pointers''' ([:GFortran43:Developed as SoC 2007 project])

Fortran 2003 Features in GNU Fortran

Implemented

See also [:GFortran#news: GFortran news] and in the [:GFortran#manuals: GNU Fortran manuals] and note that the following list is incomplete.

  • Command-line intrinsics

  • Streaming IO

  • ENUMERATION support

  • Allocatable components and dummy arguments (TR 15581) and move_alloc

  • USE statements with INTRINSIC and NON_INTRINSIC (since 2006-11-17)

  • ISO_FORTRAN_ENV module (since 2006-11-20)

  • VOLATILE attribute (since 2006-11-07)

  • VALUE attribute (since 2006-12-03)

  • IMPORT statement (since 2006-12-08)

  • PROTECTED statement (since 2006-12-10)

  • Intent for pointers (since 2007-01-05)

  • Renaming operators on the USE statement (since 2007-03-08)

Unimplemented

Features supported by several other compilers

With pending patches

  • ISO C Bindings Christopher D. Rickett (crickett at lanl dot gov) [http://gcc.gnu.org/PR25709 PR25709]

    • This feature is included in the fortran-experiments SVN branch. (See also [:Fortran2003Bind_C:Work items before it can be merged].)

Without pending patches

Other features

Note: This list is by far incomplete.

  • More allowed initialization expressions [http://gcc.gnu.org/PR29962 PR29962]

  • MIN / MAX with character arguments [http://gcc.gnu.org/PR29828 PR29828]

  • Non-initialization expr. as case-selector [http://gcc.gnu.org/PR25104 PR25104]

  • type-spec for array constructor [http://gcc.gnu.org/bugzilla/show_bug.cgi?id=27997 PR27997]

  • Type-bound procedures: PROCEDURE, GENERIC

  • Abstract derived types: ABSTRACT, DEFERRED

  • ALLOCATE specifiers SOURCE= and ERRMSG=

  • SELECTED_CHAR_KIND

  • Pointer Rank Remapping [http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29785 PR29785]

  • ASSOCIATE construction

  • Optional KIND= specifier: ACHAR, COUNT, IACHAR, ICHAR, INDEX, LBOUND, LEN, LEN_TRIM, MAXLOC, MINLOC, SCAN, SHAPE, SIZE, UBOUND, VERIFY (complete?)

  • BOZ with int(), real(), dbl(), cmpl() (incl. kind parameter) [http://gcc.gnu.org/bugzilla/show_bug.cgi?id=29471 PR29471] [http://gcc.gnu.org/bugzilla/show_bug.cgi?id=18026 PR18026]

    • This clashes with the extension to allow integer BOZ everywhere; current gfortran real(z'FFFF') is equivalent to Fortran 2003's real(int(z'FFFF',kind=maxKindNeeded)). Intel's solution is to give a warning (-stand f95) that in Fortran 2003 this would be differently interpreted. [Probably one could also give a warning for -std=f2003]. The next question is how to interprete real(BOZ), probably the same as data /realvariable/ (not supported yet); our result should match the result of other compilers.

None: Fortran2003 (last edited 2017-12-09 13:11:19 by ThomasKoenig)