should not g95 be downward compatible
in [Fortran]
|
Prev: errata list
Next: select specified columns in ch. array
From: Myron on 23 Jan 2010 22:42 I've recently been forced to change from g77 to g95, and g95 flags declaration statements which contain initialization values, such as INTEGER x/5/ to be errors. Is there a compile-time option I can use to cause this to NOT be an error? --Myron
From: dpb on 23 Jan 2010 23:07 Myron wrote: > I've recently been forced to change from g77 to g95, and > g95 flags declaration statements which contain initialization values, > such as > INTEGER x/5/ > to be errors. Is there a compile-time option I can use to > cause this to NOT be an error? Not sure about the switch but the /val/ syntax was a vendor-specific extension anyway iirc... --
From: Richard Maine on 23 Jan 2010 23:51 dpb <none(a)non.net> wrote: > Myron wrote: > > I've recently been forced to change from g77 to g95, and > > g95 flags declaration statements which contain initialization values, > > such as > > INTEGER x/5/ > > to be errors. Is there a compile-time option I can use to > > cause this to NOT be an error? > > Not sure about the switch but the /val/ syntax was a vendor-specific > extension anyway iirc... Why would g95 necessarily be compatible with every nonstandard feature of g77? It isn't as though there is any relationship between g77 and g95. G95 was a completely independent compiler - not a development based on g77. Even the term "downward compatible" is a bit misleading here in that it implies a relationship that doesn't exist. G95 accepts Fortran 77 code. I'm trying to recall if there might have been some obscure feature that it omitted. I was thinking perhaps there was one, but I forget what it was and I might misrecall anyway. But you can't count on it accepting random nonstandard features from every f77 compiler that ever was (and no, g77 doesn't have a special role - see first para above). I don't have the answer about a possible special switch either. I could try to go look it up in the docs on the g95 web page, but then I've got no particular advantage over the OP in doing that. In case he doesn't happen to know, the g95 web page is at www.g95.org. -- Richard Maine | Good judgment comes from experience; email: last name at domain . net | experience comes from bad judgment. domain: summertriangle | -- Mark Twain
From: James Van Buskirk on 24 Jan 2010 01:47 "Myron" <mcalhoun(a)ksu.edu> wrote in message news:eeb4e965-d5d2-4881-94dc-a706b56ca6f7(a)n7g2000yqb.googlegroups.com... > I've recently been forced to change from g77 to g95, and > g95 flags declaration statements which contain initialization values, > such as > INTEGER x/5/ > to be errors. Is there a compile-time option I can use to > cause this to NOT be an error? gfortran tries to be more accommodating to g77 code than does g95: C:\gfortran\clf\f77_example>type f77_example.for INTEGER X /5/ WRITE(*,*) X END C:\gfortran\clf\f77_example>gfortran f77_example.for -of77_example C:\gfortran\clf\f77_example>f77_example 5 -- write(*,*) transfer((/17.392111325966148d0,6.5794487871554595D-85, & 6.0134700243160014d-154/),(/'x'/)); end
From: Ron Shepard on 24 Jan 2010 01:47
In article <hjgh40$r8d$1(a)news.eternal-september.org>, dpb <none(a)non.net> wrote: > Myron wrote: > > I've recently been forced to change from g77 to g95, and > > g95 flags declaration statements which contain initialization values, > > such as > > INTEGER x/5/ > > to be errors. Is there a compile-time option I can use to > > cause this to NOT be an error? > > Not sure about the switch but the /val/ syntax was a vendor-specific > extension anyway iirc... There are several ways to fix this syntax error in the source code. The easiest is probably integer :: x=5 Anyone know of one of the source code converter/maintenance tools that would do this automatically? $.02 -Ron Shepard |