https://gcc.gnu.org/bugzilla/show_bug.cgi?id=120769
James K. Lowden <jklowden at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|UNCONFIRMED |RESOLVED
Resolution|--- |WONTFIX
Assignee|unassigned at gcc dot gnu.org |jklowden at gcc dot
gnu.org
--- Comment #2 from James K. Lowden <jklowden at gcc dot gnu.org> ---
What does "undo" mean here? The user uses a command-line option, and the
compiler honors it. There's nothing to undo. One can use `>>SOURCE format` to
alter the compiler's interpretation of the input.
Automatic format detection is the default. In general IMO there's no need for
a command-line option to duplicate the default. If the user wants
auto-detection, he gets that for free.
Automatic format detection can be fooled, including by files that arrive via
the -include option. The heuristic is looking for PROGRAM-ID. Unless it finds
that string, and evidence of fixed-format on that line, it defaults to free
format. Admittedly the manual could be clearer about that.
As a matter of policy,
1. whenever possible gcobol uses the same command-line options as other GCC
front ends use
2. every option has a unique effect; no option duplicates the effect of
another
3. gcobol uses the exact same option name that GnuCOBOL uses only if we're
sure we can duplicate the semantics
Sometimes these objectives are in conflict. :-/
I'm going to close this PR because I see nothing to fix. I wouldn't have
chosen the names for fixed/free format that gfortran chose, but IMO consistency
outranks taste.
