[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[netCDF #UGJ-419818]: [Fwd: Re NetCDF Build Problem: Fwd: Configure Patch]

Hi Lynton,

> I encountered big problems building the fortran netcdf4 library from
> the current snap shot (using the Lahey fujitsu compuler). It turns
> out that there is a bug in the installation that was originally
> pointed out by John Storrs (see attached email) but has yet been put
> into the disitribution. The a ticket number for this bug is in one
> of the attached emails.  Can this be corrected please?

I discussed this with Ed, and he said that what John Storrs pointed
out is a libtool bug, on your platform with the Lahey lf95 compiler.
It should be reported to the libtool developers and probably Lahey as
well.  When one of them fixes the cause for the bug, netCDF will work
for shared builds with lf95.

There are various workarounds for the problem, but not much we can do,
because the configure code that John identified is generated by
autoconf and libtool from the line 


in configure.ac.  We have control over configure.ac, but editing the 
output from the automatic generation of configure is unsustainable.

Among the possible workarounds are:

 1. Figure out the correct LDFLAGS (and possibly CFLAGS and other
    environment vars) for the lf95 compiler, and set them in your
    environment.  That would make any package that uses libtool to
    generate shared libraries work for you, not just netCDF.  With
    the correct flags, any of the command line options can be

 2. Just patch the generated configure script, as John Storrs did,
    until libtool is fixed.  I think there is even a simpler patch
    than he provided, to just replace the 3 occurrences of the line in
    the 4.1.1 configure script

       pgcc* | pgf77* | pgf90* | pgf95*)

       pgcc* | pgf77* | pgf90* | pgf95* | lf90* | lf95*)

    In the latest snapshot, this would instead be replacing the 3
    occurrences of

      pgcc* | pgf77* | pgf90* | pgf95* | pgfortran*)

       pgcc* | pgf77* | pgf90* | pgf95* | pgfortran* | lf90* | lf95*)

 3. Try to fool the configure script into thinking lf95 is really
    pgf95 by setting FC=/tmp/pgf95 and make the latter a symbolic link
    to lf95 just during the configure step.  I have no idea whether
    that would work or for how long ...


Russ Rew                                         UCAR Unidata Program
address@hidden                      http://www.unidata.ucar.edu

Ticket Details
Ticket ID: UGJ-419818
Department: Support netCDF
Priority: Normal
Status: Closed

NOTE: All email exchanges with Unidata User Support are recorded in the Unidata inquiry tracking system and then made publicly available through the web. If you do not want to have your interactions made available in this way, you must let us know in each email you send to us.