[MITgcm-support] Compile problems w/ GMRedi parameterization

[Martin Losch]

Hi David,

I am returning this to the support list, as this may be interesting to others (and also warrants for a small change in the code, maybe).

So, as I guessed earlier, your packages.conf did not contain the correct list of packages. Your packages just had "gmredi", and so only the gmredi package was included in the compilation. There appears to be a minimal set of required packages without which the code does not link. They are grouped in a "macro" gfd: mom_common mom_fluxform mom_vecinv generic_advdiff debug mdsio rw monitor

If you do not specify any packages.conf you'll get what's listed in "gfd", but no "gmredi", so your packages.conf should have these two lines:

gfd
gmredi

the macro "oceanic" contains gfd, gmredi and kpp, so you could use that also. You'll find these things in pkg/pkg_*

I think that it should not be possible to specify a packages.conf that leads to code that cannot be linked. So maybe we can do something about that (change pkg_defaults? or genmake2 to stop if this minimal set is not avaiable), but you should be fine for most cases with a packages.conf that  contains the line 
oceanic
or above

Martin

On May 20, 2010, at 3:09 AM, David Lewis wrote:

> Hi Martin,
> 
> I've attached the 'code' folder you requested.  It's my first time creating a .tar file so if there's an issue please let me know and I'll resend.  I've looked in both the manual and the MITgcm tutorial directories and that is where I copied the attached packages.conf and GMREDI_OPTIONS.h files from.  I also have the data.gmredi and data.pkg files copied in my own 'input' directory.  My goal is simply to create an idealized basin run with this parameterization...nothing to complex.  Yet. :-)
> 
> What I don't really understand is how to 'mix and match'.  Do I need other files (like the CPP_OPTIONS.h file) to run the GMRedi parameterization?  Can I compile with GMRedi as my only parameterization, or do I need to run the grid or PTRACERS parameterization as well?  I checked primarily in Chapters 3,4 and 6 of the manual but couldn't really find an answer to this. 
> 
> Anyway, thank you so much for your help with this.  I know it takes up your time and I really appreciate it.
> 
> V/r,
> David
> 
> 
> > From: Martin.Losch at awi.de
> > Date: Mon, 17 May 2010 08:55:11 +0200
> > To: mitgcm-support at mitgcm.org
> > Subject: Re: [MITgcm-support] Compile problems w/ GMRedi parameterization
> > 
> > David,
> > 
> > apparently at the link-step the compiler does not find the routines, that are in the subdirectory pkg/rw.
> > 
> > I can think of only one way for this to happen: this directory has not been included explicitly or implicitly in the packages.conf file. For more help, please send you code directory (best as an "archive", gzipped tar file).
> > 
> > There are many examples that use GM in the verification directory, e.g. most of global_ocean.*, please have a look at the corresponding code directories.
> > 
> > Martin
> > 
> > On May 16, 2010, at 7:55 PM, David Lewis wrote:
> > 
> > > Martin, all,
> > > 
> > > My apologies on the attachment. Here's a resend with the attachment in .rtf form. Thanks for informing me of this.
> > > 
> > > V/r,
> > > David
> > 
> > 
> > _______________________________________________
> > MITgcm-support mailing list
> > MITgcm-support at mitgcm.org
> > http://mitgcm.org/mailman/listinfo/mitgcm-support
> 
> The New Busy think 9 to 5 is a cute idea. Combine multiple calendars with Hotmail. Get busy. <code.tar>