Update CPPs and Documentation #329
Conversation
|
|
||
| For standalone Icepack, The CPPs are defined by the `CPPDEFS` variable in the Icepack | ||
| Makefile. They are defined | ||
| by passing the -D[CPP] to the C and Fortran compilers (ie. -DNO_I8) and this |
There was a problem hiding this comment.
Instead of 'ie.' please use 'e.g.'
I don't know the Latin, but I think i.e. means "that is," and e.g. means "for example,"
|
I agree the NOFRAC should be moved to namelist. I was thinking that might take some coordination with CICE and think it will, but I believe the Icepack upgrade can be done first. I'll update this PR with that change. Good idea. |
|
The NOFRAC cpp has been removed. The quick suite was run with 4 compilers on gordon and all results are bit-for-bit. |
| r16_kind = selected_real_kind(33,4931) | ||
| #endif |
There was a problem hiding this comment.
This looks good to me, but I have a dumb question. I'm not familiar with how selected_real_kind works, but it seems to me there might be a machine and compiler dependence here? Do we have to update this every couple of years? Sorry if I missed this in an earlier discussion.
There was a problem hiding this comment.
The NO_I8 and NO_R16 are definitely compiler dependent which means they may be required on certain machines. And this could certainly change. For instance, NO_R16 is defined for a few machine/compilers now. It's possible those are incorrect. To be honest, I'm not sure how those settings were decided. I suspect it's partly historical and they have not been reviewed for a while. The main point of these CPPs is that they provide users with a flag to turn off these features if their compiler doesn't support them. As a general rule, they should only be used when needed. And we probably should review the current settings in the current machines to check whether we really do need them where they are currently set.
| @@ -17,10 +18,18 @@ module icepack_kinds | |||
| char_len_long = 256, & | |||
| log_kind = kind(.true.), & | |||
| int_kind = selected_int_kind(6), & | |||
| #if defined (NO_I8) | |||
There was a problem hiding this comment.
Minor point, but why #if defined and not #ifdef ? I think they do the same thing but maybe we should stick to one or the other?
There was a problem hiding this comment.
In CICE, there is a mixture of #ifdef and #if defined. The NO_I8 in CICE has #if defined, so I made the icepack format match that. I could change it here, but then would probably want to change it in CICE as well. If we want to take this step, I'm happy to do it. I'm personally not that concerned about mixing the styles, although agree a single approach is nicer (but then requires diligence to maintain it).
| support certain coding features to be excluded or included during the compile. They | ||
| exist in part to support the CICE model and other applications that use Icepack. | ||
|
|
||
| For standalone Icepack, The CPPs are defined by the `CPPDEFS` variable in the Icepack |
There was a problem hiding this comment.
| For standalone Icepack, The CPPs are defined by the `CPPDEFS` variable in the Icepack | |
| For standalone Icepack, The CPPs are defined by the ``CPPDEFS`` variable in the Icepack |
| #ifdef NOFRAC | ||
| wiso_alpi = c1 | ||
| #endif |
There was a problem hiding this comment.
This is the only instance of "NOFRAC"... was this added for compiling CICE as part of another model ? (@eclare108213 you added that in 1ae0446 (Isotopes for Icepack (#307), 2020-04-06))
There was a problem hiding this comment.
According to @dabail10, this was added for debugging the isotopes and is not needed at this time.
Co-authored-by: Philippe Blain <levraiphilippeblain@gmail.com>
Co-authored-by: Philippe Blain <levraiphilippeblain@gmail.com>
Co-authored-by: Philippe Blain <levraiphilippeblain@gmail.com>
Co-authored-by: Philippe Blain <levraiphilippeblain@gmail.com>
There was a problem hiding this comment.
Looks good also.
Just a note (you might already know); you can check the "add suggestion to batch" when I make suggestions so that there is only one additional commit with all my suggestions squashed. But since we squash-merge PRs it does not really matter as the individual commits are discarded on master.
PR checklist
Update CPPs and Documentation
apcraig
Ran quick suite on 4 compilers on gordon. Should not affect standalone Icepack. https://github.com/CICE-Consortium/Test-Results/wiki/icepack_by_hash_forks#8efeb5970607894ee4286684680f4acb57614ae8
Also tested in CICE with Update CPP implementation CICE#490