[MITgcm-devel] f_basic

[Martin Losch]

Hi Jean-Michel,

a thought from an "outside": the problem with having all verification  
experiments is not so much the testing itself (which can easily be  
restricted to just a few with testreport), but the downloading.  
That's something that you don't realize when you are at MIT, where  
the download speed is acceptable, but e.g. here at AWI I wait for up  
to one hour for the entire stuff to download. Therefore I certainly  
like to MITgcm_"light" versions (verif_basic, etc.)

Martin

On Jul 7, 2006, at 4:21 PM, Jean-Michel Campin wrote:

> Hi,
>
> I propose to remove the following CVS modules:
>  MITgcm_verif_basic
>  MITgcm_verif_atmos
>  MITgcm_verif_ocean
> and to add a new one:
>  MITgcm_verif_tutorials (or MITgcm_tutorials) with the code
>     and the tutorials only (without the other verification- 
> experiments)
>
> 1) The main reason: for a new user point of view, I don't think that
> downloading only few experiments is really useful:
> -more difficult to update
> -packages are documented with an illustrative verification exp.
>  that might be missing (e.g., zonal_filt with MITgcm_verif_ocean)
> -only an expert cvs-user (not like me) knows what he will get
>  when using these modules (e.g., cvs co -P was giving me few
>  empty directories ???)
>
> 2) And for us, to test a limited number of experiments (because
>  it's too long to test all of them), we can modify testreport.
>
> 3) I would like to use "prepare_run" to reduce the size
>  of the MITgcm directory to download, by avoiding duplicated
>  files (e.g., grid-cs32 files): keep only one copy
>  (my preference, the one from a tutorial_*/input dir, if possible)
>  and use it in all the different experiments.
>  This will clarify which experiments use the same grid, and which exp.
>  use a different one (same thing for bathymetry files, forcing  
> fields ...)
>  We can certainly think of other (& better) ways to store shared  
> files,
>  but this way seems an easy first step toward a clearer situation.
>  Now, back to this cvs-modules issue, if we have plenty of experiment
>  sub-sets for downloading, changing from time to time, it's becoming
>  a nightmare to keep track of where to put a shared file.
>
> Suggestions ?
>
> Jean-Michel
> _______________________________________________
> MITgcm-devel mailing list
> MITgcm-devel at mitgcm.org
> http://mitgcm.org/mailman/listinfo/mitgcm-devel