This is the mail archive of the fortran@gcc.gnu.org mailing list for the GNU Fortran project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

#gfortran meeting 2008/11/28 - (21 CET = 20 UTC = 12 PST): gfortran-4.5; what next?


Some of you will have received notice already of tonight's meeting (21
CET = 20 UTC = 12 PST) on #gfortran:

The subject: gfortran-4.5; what next?

Tobias wrote the following, which will serve as a wish list.  We
should try to establish what we want to accomplish in 2009, the order
of priority and, if possible, identify people to do the work.  I do
not think that we should go into detail about solutions or
implementation; we can set up further meetings to do that.

Let's have a good mix of maintainers and users that have been involved
in the development of gfortran.  If I have forgotten anybody on the
direct mailing, please forgive me.  This was done over breakfast:-)

Cheers

Paul

==================================================

What are the topics?I came up of the following list, which is probably
both too long and misses important parts:

- Array descriptor
- Multiple decl per function (and via this same-file argument checks?)
- inlining of non-contained procedures
- CLASS/Polymorphic types/OOP
* Preserving library ABI backward compatibility?!?

Fixing bugs, missed optimizations (arrays, temporaries, I/O library),
adding diagnostics are also work items; I don't know whether we need to
discuss something there.

Other possible items:

- IEEE
- STDCALL
- Parameterized derived types
- F2003's I/O (real asynchronous, UTF8 leftovers, user-defined I/O)
- Associated
- Pointer assignments, scalar allocatable, proc-pointer components,
smaller F2003 left overs
- TR (the one attached) for TYPE(*), dimension(*) [and bind(C),
optional, which is trivial]
- Coarrays
- Submodules
- F2008 contiguous
- Supporting Richard's middle-end array support
- Full CPP support (w/o -cpp-legacy)
- Run-time call checking procedures (esp. those which have no explicit
interface), recursion checking.
- Run-time checking for (not) modifying loop indices and for recursive
calling of no recursive procedures
(- Run-time allocation tracing (FX's patch) - This will be posted this
weekend, Paul:-) )


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]