PING: [patch, fortran] PR 27997: Implement F2003-style array constructor with typespec

[Tobias Burnus]

Daniel Kraft wrote:
> I did run the tests through the framework, so this is not the problem. 
> I'm not sure about tab-characters in the patch; shouldn't there be 
> any?  (Are they disallowed in Fortran for indentation or something 
> like this?)
>
> What I did do was replacing every eight spaces indentation by tabs as 
> I was told this is the way to go for GCC source; or should I do this 
> only for real C sources and not for testcases?
Regarding test cases: There is no real formatting standard with regards 
to test cases, use whatever you like - as long it is valid Fortran 
(unless you test the error/warning diagnostics).

Regarding tabs in Fortran: Official Fortran characters are: A-Z (which 
are equivalent to a-z), 0-9, underscore, blank, and other special 
characters, but not the tab character. (See section 3.1 in the Fortran 
2003 standard.) Other characters are allowed in comments, character 
literals and I/O (records, edit descriptors).

Essentially all compilers support TAB as extension, especially as many 
(fixed-form) FORTRAN programmers used a <tab> to replace the six spaces 
at the beginning of a normal line.

As tabs are extensions, -pedantic warns if it sees one. By default the 
test suite uses the option "-pedantic", unless one overwrites this 
option using "{ dg-options "" }".

I think the test case should have failed if you had put it into the 
testsuite/gfortran.dg/ directory before calling "make check-gfortran". 
To run only a single test, you can use

   make check-gfortran RUNTESTFLAGS='dg.exp=mytests*.f90'

Tobias