[MITgcm-devel] streamice_init_fixed.f in streamice_ad_diff.list

[Jean-Michel Campin]

Hi Dan,

OK, I missed a "special" OpenAD feature in packages_init_fixed.F (lines 471-475)
that justifies to have streamice_init_fixed.f in ad_diff.list

We could just leave with this (TAF complains but does not seem very serious ?)
or we could move "streamice_init_fixed.f" to the local: code_oad/code_ad_diff.list
There is already a file like this in verification/halfpipe_streamice/code_oad
so it would not involve too many changes (and would be cleaner for TAF build).

Let me know what you prefer and will do it soon after.

Cheers,
Jean-Michel

On Thu, Jan 18, 2018 at 08:13:09AM +0000, Daniel Goldberg wrote:
> Hi J-M
> 
> I???m not sure as this might be borne out of improper understanding of the ad_diff.list file or it could have to do with openad which makes use of the same file I believe. I will check but I???m pretty sure it can be removed.
> 
> Dan
> 
> Sent from my iPhone. Apologies for brevity.
> 
> > On Jan 17, 2018, at 11:42 PM, Jean-Michel Campin <jmc at mit.edu> wrote:
> > 
> > Hi Dan,
> > 
> > Do you remember why you added "streamice_init_fixed.f" in
> > streamice_ad_diff.list (CVS revision 1.8, on Mar 07, 2015) ?
> > 
> > S/R STREAMICE_INIT_FIXED is called from packages_init_fixed.F
> > which is not in any *_ad_diff.list, so it's not very clear to me
> > why it's needed.
> > 
> > The other thing is that within streamice_init_fixed.f, there are
> > some S/R calls that TAF does not know about, so that we now get
> > (with recent version of TAF) this type of warning:
> >> checking subroutine ** streamice_init_fixed **
> >> *WARNING* : ad_input_code.f:XXX : call of streamice_init_diagnostics too many arguments (1) defined = 0
> >> *WARNING* : ad_input_code.f:XXX : call of streamice_init_phi too many arguments (1) defined = 0
> > 
> > Cheers,
> > Jean-Michel