| minmake is a Java/XML based platform-independent make tool.
minmake does the same job as UNIX make or Windows nmake.
By default, minmake looks for its task in the file "make.xml" of
the current directory.
Here is a complete example for a project managed with minmake:
make.xml
- make file for an application of 4 modules
mdefs.xml
- general rules for MINERVA projects
usagi.xml
- host environment declarations
make.xml observes the XML format. Syntax is as follows:
make.xml
<make [default="DefaultRule"]>
(Include)*
(Var)*
(Rule|GenRule)*
</make>
Include
<include>Filename</include>
Var
<var>
<name>VariableName</name>
(Value)*
</var>
Value
<value>SomeText</value>
Rule
<rule>
Target
(Dep)*
(Body|Comment)*
</rule>
GenRule
<genrule>
Target
(Dep)*
(Body|Comment)*
</genrule>
Target
<target>TargetName</target>
Dep
<dep>Dependency</dep>
Comment
<comment>SomeText</comment>
Body
<body
[dir="WorkingDirectory"]
[input="Filename"]
[output="Filename"]
[append="Filename"]
>SomeCommand</body>
All path references, commands, and comments may contain variables.
Variables are referenced by "$(VariableName)".
Before being used, variables must be declared in a Var section.
The variable "HOSTNAME" is predefined and gives the hostname of
the machine in which minmake was invoked. This allows for
example to define different rules or variables depending an
a specific machine, like path names for system libraries,
using <include>$(HOSTNAME).xml</include> .
When variables are referenced it is possible to modify
parts of the obtained value with the construct
"$(Variable:.orig-suffix=.new-suffix)". To substitute all
occurrances of ".min" in a variable "V" with ".mbc"
use "$(V:.min=.mbc)" .
Pathnames, commands, and comments in GenRules may contain wildcards.
In the dependency section, '*' can be used as a wildcard.
Every occurance of '*' is counted from the left to right.
With $i (i in 1-9) in the Comment or Body section
the part occupied by '*' can be referenced.
Example:
A rule to compile MINERVA files:
<genrule>
<target>*.mbc</target>
<dep>$1.min</dep>
<comment>compile $1.min</comment>
<body>$(MINERVAC) $1.min</body>
</genrule>
This defines the following:
A file with the suffix .mbc depends on a file with the
same name but suffix .min . Before compiling, minmake
shall output the comment "compile <file>.min".
Compilation itself is done with the command
"$(MINERVAC) <Filename>.min",
wheren "MINERVAC" is a variable that has been defines before.
The individual contructs:
The section var defines variables.
e.g.:
<var>
<name>JAVA</name>
<value>jview d:/nologo=true /cp:a</value>
</var>
<var>
<name>MINERVA_HOME</name>
<value>c:\minerva</value>
</var>
<var>
<name>MINERVAC</name>
<value>$(JAVA) $(MINERVA_HOME) com.ifcomputer.minerva.Application -c $(MINERVA_HOME)/minerva.mca</value>
</var>
Variables may have more than one value:
<var>
<name>SOURCE</name>
<value>file-1.min</value>
<value>file-2.min</value>
<value>file-3.min</value>
</var>
In this case the value of the variable is the concatination
of the invidual value fields with a space character.
For the example above this results in SOURCE="file-1.min file-2.min file-3.min".
The section Rule defines rules. These rule are processed when
needed by minmake (as known from traditional make).
The section Rule can only give rules for specific files, i.e.
there are no wildcards allowed in the target part.
Rule does not require a body part.
In that case, depedency rules are defined.
If there is both a GenRule and a Rule, then only the more specific
Rule is processed. GenRules and Rules with only dep sections (i.e.
without body sections) are always processed.
Rules may optionally have multiple Attributes.
The attribute "input" requests that the given file be used as
standard input for the command.
The attribute "output"requests that the given file be used as
standard output for the command.
The attribute "append" requests that the standard output be added
to the given file.
The attribute "dir" requests that the command be executed in the
given directory. This is important for recursive make files.
e.g.
<rule>
<target>all</target>
<body dir="lib">$(MAKE) all</body>
<body dir="kernel">$(MAKE) all</body>
<body dir="bips">$(MAKE) all</body>
</rule>
So this command only walks into the directories bin, kernel, lib
respectively and executes make there.
minmake accepts the following command line parameters:
-f <Filename> load the given minmake-file (default: make.xml)
Name=Value assing value Value to the variable Variable
Goal Rule to be activated. Default is determined by the
attibute "default" of <make>.
minmake may be called with multiple goals.
If no goal is given, the the goal given in the
attribute "default" is processed.
If this is not defined and no goal given explicitly,
then the goal "all" is processed.
minmake knows a few predefined commands. Predefined commands can
be used just as normal commands. The advantage of predefined
commands is only to hide operating system dependencies.
The following commands are predefined:
del <file> ...
delete all given files, recursively if directories
copy <file> ... <directory>
copy files into a directpry, recursively if directories
copy <file1> <file2>
copy file1 to file2
move <file1> <file2>
rename file1 to file2
mkdir <directory> ...
make directies if non existant.
All needed intermediate directories are also made.
(e.g. "mkdir a/b/c" suffices instead of "mkdir a a/b a/b/c"
echo text
output some text
cat <file> ...
all given files are written in sequence to standard output.
To redirect into another file use the attribute "output"
or "append"
e.g. <body output="res.txt">cat file1.txt file2.txt file3.txt</body>
Except for "cat" all predefined commands ignore the attributes
of the body construct. All predefined commands ignore the "dir"
attribute.
| 
