This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC 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]

Re: [patch,fortran] Allow namelist with internal file (PR 28224)


Tobias Burnus wrote:
Hi,

Paul Thomas schrieb:
Might it not be better to use -std=f95 for this test....
* gfortran.dg/namelist_internal.f90: New namelist test.
* gfortran.dg/namelist_internal2.f90: New namelist test.
...and to drop this latter?
Well, the first test test whether internal file read/write of namelists
works.
The second test tests whether it is still forbidden for -std=f95.
The latter means the last one...
I can drop either test, but I cannot simply change -std=f2003 to
-std=f95 in the first test; doing all the modifications needed, I will
end up with the second test.
Test the new feature but try to keep the old all inclusive test is what I was after.

Paul
That is anyhow the question: How much should one test? Only a minimal
set, which will get easily broken or also more extensively the behaviour
of a function, even if it is unlikely that a regression occurs?
The former prevents bloating, the latter reduces the chance of more
unlikely regressions.

In this regard, shall I check in namelist_internal.f90 or
namelist_internal2.f90 or both?
(If only namelist_internal2.f90, I will rename it to namelist_internal.f90.)

Tobias




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