siili@finsun.csc.fi (Tero Siili) (06/10/91)
I am doing some preliminary testing on a Sun 4 (a Cray front-end) and am using the Sun F77 compiler. I would need to get a compilation listing with variable and constant cross-reference listing etc. For some reason I have been unable to find the suitable switch to accomplish that. What am I missing? How do I create the compilation listing (.l) file? Tero Siili Finnish Meteorological Institute
khb@chiba.Eng.Sun.COM (Keith Bierman fpgroup) (06/11/91)
In article <SIILI.91Jun10102650@finsun.csc.fi> siili@finsun.csc.fi (Tero Siili) writes:
...
with variable and constant cross-reference listing etc. For some reason
I have been unable to find the suitable switch to accomplish that.
...
Because there is no such switch. An online/interactive tool called the
sourcebrowser is part of the package, compile -sb and check out man
sbrowser.
Cross references are generated by some of the "lint" tools available,
notably IPT's fortran lint and QUIBUS's forwarn.
--
----------------------------------------------------------------
Keith H. Bierman keith.bierman@Sun.COM| khb@chiba.Eng.Sun.COM
SMI 2550 Garcia 12-33 | (415 336 2648)
Mountain View, CA 94043john@spectre.unm.edu (John Prentice) (06/11/91)
In article <KHB.91Jun10160213@chiba.Eng.Sun.COM> khb@chiba.Eng.Sun.COM (Keith Bierman fpgroup) writes: > >Because there is no such switch. An online/interactive tool called the >sourcebrowser is part of the package, compile -sb and check out man >sbrowser. > >Cross references are generated by some of the "lint" tools available, >notably IPT's fortran lint and QUIBUS's forwarn. Why is this? The UNIX Fortran compilers are the only ones I have ever encountered that don't allow you to generate cross-reference and variable listings while compiling (I should be more specific - the non-supercomputer UNIX Fortran compilers). These are very valuable debugging tools, I really miss then. Certainly dbx and dbxtool don't make up for them (and these are not very impressive debuggers anyway). I appreciate that tools exist out there to do this, but who wants to have to run so many different utilities (but that is a common complaint about UNIX - this is defintely an operating system for people who grove on lots of little utilities). Just curious. John -- John K. Prentice john@spectre.unm.edu (Internet) Computational Physics Group Amparo Corporation, Albuquerque, NM
khb@chiba.Eng.Sun.COM (Keith Bierman fpgroup) (06/11/91)
In article <1991Jun10.234931.5147@ariel.unm.edu> john@spectre.unm.edu (John Prentice) writes:
Why is this? The UNIX Fortran compilers are the only ones I have
The cannonical Unix FORTRAN compiler, sif's f77 didn't include one.
In many ways, the Unix mindset was defined by the original C compiler
and affliated tools. One can argue (as other systems implemented) that
compilers should do all sorts of static correctness analysis, cross
reference generation and other goodies. However, the Unix
tradition/mindset is one task==one tool.
miss then. Certainly dbx and dbxtool don't make up for them (and
sourcebrowser is a very different tool than dbx.
an operating system for people who grove on lots of little utilities).
Yes. It's proponents count that as a feature (combined with the usual
arguments about the joys of shells, piping, and empowering programmers
to build what they want).
I don't happen to subscribe to such ideals, for whatever that's worth.
--
----------------------------------------------------------------
Keith H. Bierman keith.bierman@Sun.COM| khb@chiba.Eng.Sun.COM
SMI 2550 Garcia 12-33 | (415 336 2648)
Mountain View, CA 94043john@spectre.unm.edu (John Prentice) (06/11/91)
In article <KHB.91Jun10220038@chiba.Eng.Sun.COM> khb@chiba.Eng.Sun.COM (Keith Bierman fpgroup) writes: > > an operating system for people who grove on lots of little utilities). > >Yes. It's proponents count that as a feature (combined with the usual >arguments about the joys of shells, piping, and empowering programmers >to build what they want). > >I don't happen to subscribe to such ideals, for whatever that's worth. > I have to agree with Keith here, I don't care much for the UNIX approach to little utilities either. Coming from either the supercomputer or PC worlds, I feel like I have taken a giant step into the past everytime I encounter UNIX. This must have been a hot operating system in the early 70's, but it leaves alot to be desired as a modern operating system. Looking at it from the supercomputer perspective, it is just not as efficient as other operating systems like CTSS. It is also less functional as a rule. All I can say UNIX has over CTSS is standardized mediocrity. Comparing it to PC's, UNIX as an operating system is certainly an improvement over MS-DOS, but the utilities are a bloody joke. The word processor on my old XT is better than almost anything I can get on UNIX (this is slowly changing. Now if only OpenWindows could manage to keep a window up reliably!). This old UNIX hacker philosophy of small utilties is in my opinion an anachronism, more appropiate to the small memory slow machines of the past than to the modern workstation or supercomputer. I would dearly love to see it discarded. Along those lines, I would love to see Fortran compilers which provide all the features that I now have to run half a dozen UNIX utilities to get. I get them on the Cray, I wish I could get them on the more user friendly (or at least less expensive) environment of my workstation. Arghhh.... John -- John K. Prentice john@spectre.unm.edu (Internet) Computational Physics Group Amparo Corporation, Albuquerque, NM
ercs50@castle.ed.ac.uk (simon) (06/13/91)
In article <1991Jun10.234931.5147@ariel.unm.edu> john@spectre.unm.edu (John Prentice) writes: > >Why is this? The UNIX Fortran compilers are the only ones I have ever >encountered that don't allow you to generate cross-reference and variable >listings while compiling (I should be more specific - the non-supercomputer >UNIX Fortran compilers). These are very valuable debugging tools, I really >miss then. Certainly dbx and dbxtool don't make up for them (and these >are not very impressive debuggers anyway). I appreciate that tools exist >out there to do this, but who wants to have to run so many different >utilities (but that is a common complaint about UNIX - this is defintely This is not all strictly true. There are *some* Unix compilers which do provide cross reference listings, post mortem variable dumps and parameter checking across calls and still support the standard debuggers like dbx and sdb. Edinburgh Portable Compilers' epcf77 for Unix systems running on SPARC,Intel, M680X0 and M88X00 is an example of such a Fortran compiler. ----------- Simon Lawrence (simon@uk.ac.ed.epcfta) Edinburgh Portable Compilers Ltd. Edinburgh, Scotland