Syntax errors in physics routines: *- and *+

[davegill]

TYPE: bug fix

KEYWORDS: HPE, CCE, syntax

SOURCE: Tricia Balle (HPE)

DESCRIPTION OF CHANGES:
Problem:
There were three types of syntax errors that were uncovered by the HPE CCE compiler.

1. Open statement used action="append".
2. Multiple unary operators next to each other *+
3. Multiple unary operators next to each other *-1.

Solution:

1. The correct usage is position="append".
2. The developer agrees with the new modification to GS3*+SASR1 was the original intent.
3. Typically, *(-1) is intended, and this is the case.

Via some grep'ing, no other occurrences of *+ or *- exist in the physics directory in the compilable
source code. Syntactically OK source code with /- (such as for DATA statements) were the only compilable
examples found. So, we fixed the above problems identified in this PR, and we could not find any other
similar examples in the source code.

LIST OF MODIFIED FILES:
modified: phys/module_mp_nssl_2mom.F
modified: phys/module_mp_ntu.F
modified: phys/module_sf_noahmplsm.F

TESTS CONDUCTED:

1. These fixes allow the WRF code to build with the CCE compiler.
2. Jenkins tests are all PASS.
3. One of the errors was a typo in the sink term of aggregates number from hail. A quick evaluation of its impact was
   quite minor from the 2D idealized simulation (the figure is not shown).

RELEASE NOTE: A few syntax errors in NSSL, NTU, an NoahMP were fixed. Most compilers skipped over them. No impact for users for whom the code already compiled.

[davegill]

@weiwangncar @dudhia
Folks,
The modification to the NTU scheme needs to be reviewed. I have no idea what was intended. This probably needs to get back to the developer.

[davegill]

@weiwangncar @dudhia
This one is the biggy

[weiwangncar]

@davegill This looks good to me. I'll ask the author to approve it too.

[davegill]

Thanks Wei.
Could we also ask the author to comment on the likely impact of multiplying, instead of adding those terms? That might be significant.

[weiwangncar]

@davegill I sent him this page, and hopefully he will see your comment.

[davegill]

jenkins is OK

Please find result of the WRF regression test cases in the attachment. This build is for Commit ID: e781e2e960665840b89ca2ce156063f4cd4c1282, requested by: davegill for PR: https://github.com/wrf-model/WRF/pull/1577. For any query please send e-mail to David Gill.

    Test Type              | Expected  | Received |  Failed
    = = = = = = = = = = = = = = = = = = = = = = = =  = = = =
    Number of Tests        : 19           17
    Number of Builds       : 48           45
    Number of Simulations  : 160           155        0
    Number of Comparisons  : 101           100        0

    Failed Simulations are: 
    None
    Which comparisons are not bit-for-bit: 
    None

[tzuchin12]

Hi, thanks. It is a typo in the sink term of aggregates number from hail. The quick evaluation of its impact is quite minor from the 2D idealized simulation (the figure is not shown).

[milancurcic]

Open statement used action="append".

FWIW, we just hit this with the IBM compiler (xlf).

Errors like this can be caught by CI, for example by disabling non-standard extensions with gfortran:

$ gfortran -std=f2018 test.f90 
test.f90:1:41:

    1 | open(20, file='new.txt', access='append')
      |                                         1
Error: GNU Extension: ACCESS specifier in OPEN statement at (1) has value ‘APPEND’

I tried looking for a similar option with the Intel compiler and couldn't find it. At least ifort -stand f18 -standard-semantics -warn all doesn't catch this.