| 2 |
% $Name$ |
% $Name$ |
| 3 |
|
|
| 4 |
\section{Using MITgcm Packages} |
\section{Using MITgcm Packages} |
| 5 |
|
\label{sec:pkg:using} |
| 6 |
|
\begin{rawhtml} |
| 7 |
|
<!-- CMIREDIR:package_using: --> |
| 8 |
|
\end{rawhtml} |
| 9 |
|
|
| 10 |
The set of packages that will be used within a partiucular model can |
The set of packages that will be used within a partiucular model can |
| 11 |
be configured using a combination of both ``compile--time'' and |
be configured using a combination of both ``compile--time'' and |
| 67 |
|
|
| 68 |
For run--time package control, MITgcm uses flags set through a |
For run--time package control, MITgcm uses flags set through a |
| 69 |
\texttt{data.pkg} file. While some packages (\textit{eg.} |
\texttt{data.pkg} file. While some packages (\textit{eg.} |
| 70 |
\texttt{debug}, \texttt{mnc}, \texttt{cost}) may have their own |
\texttt{debug}, \texttt{mnc}, \texttt{exch2}) may have their own usage |
| 71 |
conventions, most follow a simple flag naming convention of the |
conventions, most follow a simple flag naming convention of the form: |
|
form: |
|
| 72 |
\begin{verbatim} |
\begin{verbatim} |
| 73 |
usePackageName=.TRUE. |
usePackageName=.TRUE. |
| 74 |
\end{verbatim} |
\end{verbatim} |
| 75 |
where the \texttt{usePackageName} variable can activate or disable the |
where the \texttt{usePackageName} variable can activate or disable the |
| 76 |
package at runtime. |
package at runtime. As mentioned previously, packages must be |
| 77 |
|
included in order to be activated. Generally, such mistakes will be |
| 78 |
|
detected and reported as errors by the code. However, users should |
| 79 |
|
still be aware of the dependency. |
| 80 |
|
|
| 81 |
|
|
| 82 |
%\subsection{Modifying or Creating Packages} |
\section{Package Coding Standards} |
| 83 |
|
|
| 84 |
|
The following sections describe how to modify and/or create new MITgcm |
| 85 |
|
packages. |
| 86 |
|
|
| 87 |
|
\subsection{Packages are Not Libraries} |
| 88 |
|
|
| 89 |
|
To a beginner, the MITgcm packages may resemble libraries as used in |
| 90 |
|
myriad software projects. While future versions are likely to |
| 91 |
|
implement packages as libraries (perhaps using FORTRAN90/95 syntax) |
| 92 |
|
the current packages (FORTRAN77) are \textbf{not} based upon any |
| 93 |
|
concept of libraries. |
| 94 |
|
|
| 95 |
|
\subsubsection{File Inclusion Rules} |
| 96 |
|
|
| 97 |
|
Instead, packages should be viewed only as directories containing |
| 98 |
|
``sets of source files'' that are built using some simple mechanisms |
| 99 |
|
provided by \texttt{genmake2}. Conceptually, the build process adds |
| 100 |
|
files as they are found and proceeds according to the following rules: |
| 101 |
|
\begin{enumerate} |
| 102 |
|
\item \texttt{genmake2} locates a ``core'' or main set of source files |
| 103 |
|
(the \texttt{-standarddirs} option sets these locations and the |
| 104 |
|
default value contains the directories \texttt{eesupp} and |
| 105 |
|
\texttt{model}). |
| 106 |
|
|
| 107 |
|
\item \texttt{genmake2} then finds additional source files by |
| 108 |
|
inspecting the contents of each of the package directories: |
| 109 |
|
\begin{enumerate} |
| 110 |
|
\item As the new files are found, they are added to a list of source |
| 111 |
|
files. |
| 112 |
|
|
| 113 |
|
\item If there is a file name ``collision'' (that is, if one of the |
| 114 |
|
files in a package has the same name as one of the files |
| 115 |
|
previously encountered) then the file within the newer (more |
| 116 |
|
recently visited) package will superseed (or ``hide'') any |
| 117 |
|
previous file(s) with the same name. |
| 118 |
|
|
| 119 |
|
\item Packages are visited (and thus files discovered) {\it in the |
| 120 |
|
order that the packages are enabled} within \texttt{genmake2}. |
| 121 |
|
Thus, the files in \texttt{PackB} may superseed the files in |
| 122 |
|
\texttt{PackA} if \texttt{PackA} is enabled before \texttt{PackB}. |
| 123 |
|
Thus, package ordering can be significant! For this reason, |
| 124 |
|
\texttt{genmake2} honors the order in which packages are |
| 125 |
|
specified. |
| 126 |
|
\end{enumerate} |
| 127 |
|
\end{enumerate} |
| 128 |
|
|
| 129 |
|
These rules were adopted since they provide a relatively simple means |
| 130 |
|
for rapidly including (or ``hiding'') existing files with modified |
| 131 |
|
versions. |
| 132 |
|
|
| 133 |
|
\subsubsection{Conditional Compilation and \texttt{PACKAGES\_CONFIG.h}} |
| 134 |
|
|
| 135 |
|
Given that packages are simply groups of files that may be added or |
| 136 |
|
removed to form a whole, one may wonder how linking (that is, FORTRAN |
| 137 |
|
symbol resolution) is handled. This is the second way that |
| 138 |
|
\texttt{genmake2} supports the concept of packages. Basically, |
| 139 |
|
\texttt{genmake2} creates a \texttt{Makefile} that, in turn, is able |
| 140 |
|
to create a file called \texttt{PACKAGES\_CONFIG.h} that contains a set |
| 141 |
|
of C pre-processor (or ``CPP'') directives such as: |
| 142 |
|
\begin{verbatim} |
| 143 |
|
#undef ALLOW_KPP |
| 144 |
|
#undef ALLOW_LAND |
| 145 |
|
... |
| 146 |
|
#define ALLOW_GENERIC_ADVDIFF |
| 147 |
|
#define ALLOW_MDSIO |
| 148 |
|
... |
| 149 |
|
\end{verbatim} |
| 150 |
|
These CPP symbols are then used throughout the code to conditionally |
| 151 |
|
isolate variable definitions, function calls, or any other code that |
| 152 |
|
depends upon the presence or absence of any particular package. |
| 153 |
|
|
| 154 |
|
An example illustrating the use of these defines is: |
| 155 |
|
\begin{verbatim} |
| 156 |
|
#ifdef ALLOW_GMREDI |
| 157 |
|
IF (useGMRedi) CALL GMREDI_CALC_DIFF( |
| 158 |
|
I bi,bj,iMin,iMax,jMin,jMax,K, |
| 159 |
|
I maskUp, |
| 160 |
|
O KappaRT,KappaRS, |
| 161 |
|
I myThid) |
| 162 |
|
#endif |
| 163 |
|
\end{verbatim} |
| 164 |
|
which is included from the file |
| 165 |
|
\filelink{calc\_diffusivity.F}{model-src-calc_diffusivity.F} |
| 166 |
|
and shows how both the compile--time \texttt{ALLOW\_GMREDI} flag and the |
| 167 |
|
run--time \texttt{useGMRedi} are nested. |
| 168 |
|
|
| 169 |
|
There are some benefits to using the technique described here. The |
| 170 |
|
first is that code snippets or subroutines associated with packages |
| 171 |
|
can be placed or called from almost anywhere else within the code. |
| 172 |
|
The second benefit is related to memory footprint and performance. |
| 173 |
|
Since unused code can be removed, there is no performance penalty due |
| 174 |
|
to unnecessary memory allocation, unused function calls, or extra |
| 175 |
|
run-time \texttt{IF (...)} conditions. The major problems with this |
| 176 |
|
approach are the potentially difficult-to-read and difficult-to-debug |
| 177 |
|
code caused by an overuse of CPP statements. So while it can be done, |
| 178 |
|
developers should exerecise some discipline and avoid unnecesarily |
| 179 |
|
``smearing'' their package implementation details across numerous |
| 180 |
|
files. |
| 181 |
|
|
| 182 |
|
|
| 183 |
|
\subsubsection{Package Startup or Boot Sequence} |
| 184 |
|
|
| 185 |
|
Calls to package routines within the core code timestepping loop can |
| 186 |
|
vary. However, all packages should follow a required "boot" sequence |
| 187 |
|
outlined here: |
| 188 |
|
|
| 189 |
|
{\footnotesize |
| 190 |
|
\begin{verbatim} |
| 191 |
|
1. S/R PACKAGES_BOOT() |
| 192 |
|
: |
| 193 |
|
CALL OPEN_COPY_DATA_FILE( 'data.pkg', 'PACKAGES_BOOT', ... ) |
| 194 |
|
|
| 195 |
|
|
| 196 |
|
2. S/R PACKAGES_READPARMS() |
| 197 |
|
: |
| 198 |
|
#ifdef ALLOW_${PKG} |
| 199 |
|
if ( use${Pkg} ) |
| 200 |
|
& CALL ${PKG}_READPARMS( retCode ) |
| 201 |
|
#endif |
| 202 |
|
|
| 203 |
|
3. S/R PACKAGES_INIT_FIXED() |
| 204 |
|
: |
| 205 |
|
#ifdef ALLOW_${PKG} |
| 206 |
|
if ( use${Pkg} ) |
| 207 |
|
& CALL ${PKG}_INIT_FIXED( retCode ) |
| 208 |
|
#endif |
| 209 |
|
|
| 210 |
|
4. S/R PACKAGES_CHECK() |
| 211 |
|
: |
| 212 |
|
#ifdef ALLOW_${PKG} |
| 213 |
|
if ( use${Pkg} ) |
| 214 |
|
& CALL ${PKG}_CHECK( retCode ) |
| 215 |
|
#else |
| 216 |
|
if ( use${Pkg} ) |
| 217 |
|
& CALL PACKAGES_CHECK_ERROR('${PKG}') |
| 218 |
|
#endif |
| 219 |
|
|
| 220 |
|
5. S/R PACKAGES_INIT_VARIABLES() |
| 221 |
|
: |
| 222 |
|
#ifdef ALLOW_${PKG} |
| 223 |
|
if ( use${Pkg} ) |
| 224 |
|
& CALL ${PKG}_INIT_VARIA( ) |
| 225 |
|
#endif |
| 226 |
|
|
| 227 |
|
6. S/R DO_THE_MODEL_IO |
| 228 |
|
|
| 229 |
|
#ifdef ALLOW_${PKG} |
| 230 |
|
if ( use${Pkg} ) |
| 231 |
|
& CALL ${PKG}_DIAGS( ) |
| 232 |
|
#endif |
| 233 |
|
|
| 234 |
|
7. S/R PACKAGES_WRITE_PICKUP() |
| 235 |
|
|
| 236 |
|
#ifdef ALLOW_${PKG} |
| 237 |
|
if ( use${Pkg} ) |
| 238 |
|
& CALL ${PKG}_WRITE_PICKUP( ) |
| 239 |
|
#endif\end{verbatim} |
| 240 |
|
} |
| 241 |
|
|
| 242 |
|
|
| 243 |
|
\subsubsection{Adding a package to PARAMS.h and packages\_boot()} |
| 244 |
|
|
| 245 |
|
An MITgcm package directory contains all the code needed for that package apart |
| 246 |
|
from one variable for each package. This variable is the {\tt use\$\{Pkg\} } |
| 247 |
|
flag. This flag, which is of type logical, {\bf must} be declared in the |
| 248 |
|
shared header file {\it PARAMS.h} in the {\it PARM\_PACKAGES} block. This |
| 249 |
|
convention is used to support a single runtime control file {\it data.pkg} |
| 250 |
|
which is read by the startup routine {\it packages\_boot()} and that sets a |
| 251 |
|
flag controlling the runtime use of a package. This routine needs to be able to |
| 252 |
|
read the flags for packages that were not built at compile time. Therefore |
| 253 |
|
when adding a new package, in addition to creating the per-package directory |
| 254 |
|
in the {\it pkg/} subdirectory a developer should add a {\tt use\$\{Pkg\} } |
| 255 |
|
flag to {\it PARAMS.h} and a {\tt use\$\{Pkg\} } entry to the |
| 256 |
|
{\it packages\_boot()} {\it PACKAGES} namelist. |
| 257 |
|
The only other package specific code that should appear outside the individual |
| 258 |
|
package directory are calls to the specific package API. |
| 259 |
|
|
| 260 |
|
|
Parent Directory
| 
Revision Log
| 
Revision Graph
| 
Patch