Gfortran 4.3 projects

[Tobias Burnus]

François-Xavier Coudert schrieb:
>> Possible jobs:
> I would add PR 29121: CPU_TIME subroutine missing for REAL(kind=10)
> argument
I have a question in similar spirit (but to PR28484; support real-kind
count_rate):

   SYSTEM_CLOCK ([COUNT, COUNT_RATE, COUNT_MAX])

allows different kinds of integers (4,8) for count, count_rate,
count_max and count_rate might even be real (4,8,10,16). If one now
starts mixing, how should one handle this in libgfortran?

Currently, two versions exists:
system_clock_4(GFC_INTEGER_4 *count, GFC_INTEGER_4 *count_rate,
               GFC_INTEGER_4 *count_max)

system_clock_8 (GFC_INTEGER_8 *count, GFC_INTEGER_8 *count_rate,
                GFC_INTEGER_8 *count_max)

But taking all combinations into account, there are 24 possibilities;
should one assume same integer kind and only real(4) and real(8)? Then
there are only two additional cases.

Tobias