19991207: dcgrib problem under linux

  This message is in MIME format.  The first part should be readable text,
  while the remaining parts are likely unreadable without MIME-aware tools.
  Send mail to address@hidden for more info.


Pete,

The problem with MAIN__ seems like it is using the fort77 
as the linker instead of gcc.

I notice that your get a compiler warning at line 423 of put_gem_grid.f
for the st_inch call, which is obviously different from what
I am currently running. I will attatch my current Linux dcgrib to
this email message which hasn't been stripped.

If the decoder still hangs, issue a "kill -6 pid" to the dcgrib
process ID from another window so that a core will be dumped.
That should allow you to use gdb to find where the decoder
is getting stuck in the loop. The dcgrib I have is statically linked
(eg uses libf2c.a as defined in Makeinc.linux) so it should 
be able to trace using gdb.

Steve Chiswell
Unidata User Support


On Tue, 7 Dec 1999, Pete Pokrandt wrote:

> 
> Steve and all,
> 
> I am having a problem with the dcgrib decoder under linux. We are trying
> to convert grib files written by Greg Tripoli's UW-NMS model, and
> while the GRIB files look correct, and in fact can be read by various
> other grib decoders under linux (wxp can read the files just fine, for
> example), dcgrib gets through about the first 10 or so grib records
> and just hangs, not writing any more data to the gempak file, and not
> crashing. It will run endlessly for days, so it must be somehow stuck in
> an endless loop somewhere.
> 
> The exact same grib files decode just fine using dcgrib on an SGI
> workstation (under both SGI and Linux I used Gempak 5.4 pl12 decoders)
> 
> I'd like to dig into it further on my own, but I remember you saying
> there were lots of hoops you had to jump through to get gempak to
> compile under linux. I did try compiling dcgrib on my own, using the
> Makefile included, and the precompiled gempak libraries from the gbuddy
> account, but upon loading, I get an "undefinied reference to MAIN__" at
> link time, apparently coming from my libf2c library. The output from
> my compilation is included below, in case it's of any use.
> 
> Is there any (relatively easy) way I can compile just the dcgrib
> program on my own to throw in some debugging code to try to figure
> out where it is hanging, or would you have some time to take a look
> if I provided you the grib file?
> 
> I suspect that something strange is going on in the byte swapping
> that is needed on the intel platform (based on the fact that it
> works fine on an SGI)
> 
> Thanks.
> 
> Pete
> 
> Here's my compile output
> 
> [gempak@ProfHorn dcgrib]$ make dcgrib
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ dcgrib.c
> fort77 -g   -c bd_gem_grid.f
>    bd_gem_grid:
>    BLOCK DATA bd_gem_grid:
> fort77 -g   -c open_gem_grid.f
>    open_gem_grid:
> Warning on line 167: local variable ermiss never used
> fort77 -g   -c put_gem_grid.f
>    put_gem_grid:
> Warning on line 423: inconsistent calling sequences for st_inch,
>         arg 2: here real variable, previously character variable.
> Warning on line 423: inconsistent calling sequences for st_inch:
>         here 3, previously 4 args and string lengths.
> Warning on line 623: local variable ermiss never used
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ gempds.c
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ stagger.c
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ emalloc.c
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ gbds.c
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ gbytem.c
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ gdes.c
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ get_prod.c
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ grib1.c
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ gribtypes.c
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ mkdirs_open.c
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ params.c
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ product_data.c
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ quasi.c
> gcc  -DUNDERSCORE -DLINUX -DULTRIX -g   -c -DGEMPAK -D__STDC__ inetutil.c
> /usr/lib/libf2c.so: undefined reference to `MAIN__'
> make: *** [dcgrib] Error 1
> 
> 
> --
> +>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>+<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<+
> ^ Pete Pokrandt                    V 1447  AOSS Bldg  1225 W Dayton St^
> ^ Systems Programmer               V Madison,         WI     53706    ^
> ^                                  V      address@hidden       ^
> ^ Dept of Atmos & Oceanic Sciences V (608) 262-3086 (W)  222-0919 (H) ^
> ^ University of Wisconsin-Madison  V       262-0166 (Fax)262-3086 (VM)^
> +<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<+>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>+
> 
>