One of the major components new to Sencha Cmd with version 3 is the compiler. In a nutshell,
the compiler is a JavaScript-to-JavaScript, framework-aware optimizer. It is designed to
"understand" your high-level Ext JS and Sencha Touch code and produce the smallest, most
efficient code possible to support these high-level abstractions.
{@img ../command/sencha-command-128.png}
Before using the compiler, you should understand the basics of Sencha Cmd by reading the
following guides:
Under the covers, the compiler manages a set of source files and analyzes these files to
determine their dependencies. The set of all files is determined by the classpath
:
sencha compile -classpath=sdk/src,app ...
In this example, the compiler recursively loads "*.js"
from the specified list of folders.
This set of all files defines the basis for all operations to follow (that is, it defines
the "universe").
The default classpath used by the compiler comes from these configuration properties:
${framework.classpath},${workspace.classpath},${app.classpath}
The compiler's output commands (for example, concat
and metadata
) operate on the set
of files called the "current set". The current set starts out equal to the universe of all
files, but this can be manipulated using the many commands provided to perform set
operations.
Note. With the compiler, you will often see rather long command lines using the command
chaining mechanism and
. Also, in practical use cases, for long command lines you should
consider using Ant or a "response file". See
Advanced Sencha Cmd. In this guide, all command lines will be
complete (and potentially long) to keep the examples as clear as possible.
concat
A compiler ultimately is all about writing useful output given some number of inputs. Theconcat
command is designed to concatenate the source for the current set of files in the
appropriate dependency order.
The one required parameter is -out
, which indicates the name of the output file. There
are other options, however, that effect the generated file. You can pick one of the
following options for compression:
-compress
- Compress the generated file using the default compressor. Currently this-yui
.-max
- Compress the generated file using all compressors and keep the smallest.-closure
- Compress the generated file using Google Closure Compiler.-uglify
- Compress the generated file using UglifyJS.-yui
- Compress the source file using YUI Compressor.-strip
- Strip comments from the output file, but preserve whitespace. This is theThe following command illustrates how to produce three flavors of output given a single
read of the source.
sencha compile -classpath=sdk/src \
exclude -namespace Ext.chart and \
concat ext-all-nocharts-debug-w-comments.js and \
-debug=true \
concat -strip ext-all-nocharts-debug.js and \
-debug=false \
concat -yui ext-all-nocharts.js
The compiler can also generate metadata in many useful ways, for example, the names of
all source files, the set of files in dependency order, etc. To see what is available,
see the Generating Metadata guide.
When you need to produce multiple output files, it can be very helpful to save the
current set for later use, which you do like this:
sencha compile -classpath=sdk/src \
exclude -namespace Ext.chart and \
save nocharts and \
...
restore nocharts and \
...
The
savecommand takes a snapshot of the current set and stores it under the given name (
nocharts` in the above).
The simplest use of a saved set is the restore
command. This does the reverse and
restores the current set to its state at the time of the save
.
Many of the commands provided by the compiler are classified as set operations, which are
operations that work on and produce sets. In the case of the compiler, this means sets of
files or classes. Let's first take a look at set terminology.
There are three classic set operations:
Intersection - The intersection of two sets is a set containing only what was in both
sets.
{@img set-intersect.png}
Union - The union of two sets is a set containing whatever was in either of the sets.
{@img set-union.png}
Difference - The difference of two sets is the set of all things in the first set that
are not in the second set.
{@img set-difference.png}
include
and exclude
These two set operations are probably the most common (and flexible) set operations. Both
support these basic switches:
-namespace
- Matches files that define types in the specified namespace.-class
- Matches a specific defined type.-file
- Matches filenames and/or folder names using Ant-style glob patterns (a "*"-tag
- Matches any files with the specified tag(s) (see below).-set
- The files that are present in any of the specified named sets.In all of these cases, the next command line argument is a list of match criteria
separated by commas. Also, a single exclude
or include
can have as many switch/value
pairs as needed.
So, let's start with a simple example and build an "ext-all-no-charts-debug-w-comments.js"
.
sencha compile -classpath=sdk/src \
exclude -namespace Ext.chart and \
...
What is happening here is that we started with only the Ext JS sources (in "sdk/src"
) and
they were all part of the "current set". We then performed a set difference by excluding
all files in the Ext.chart
namespace. The current set was then equivalent to "ext-all.js"
but without any of the Chart package.
include
and exclude
with -not
Both include
and exclude
support a rich set of matching criteria. This is rounded out
by the -not
switch, which negates the matching criteria that follows it. This means that
the files included or excluded are all those that do not match the criteria.
For example:
sencha compile -classpath=sdk/src,js \
... \
exclude -not -namespace Ext and \
...
The above exclude
command will exclude from the current set any classes that are not in
the Ext
namespace.
all
SetIn some cases, it is very handy to restore the current set to all files or to the empty
set. To do this, you simply use include
or exclude
with the -all
switch. To build
on the previous example:
sencha compile -classpath=sdk/src \
... \
include -all and \
... \
exclude -all and \
...
After the include -all
, the current set is all files. After exclude -all
it is the
empty set.
As shown already, the include
command is a form of set union: it performs a union of
the current set with the set of matching files. Sometimes it is desirable to not include
the current set in the union and only those file matching the desired criteria. This is
what the union
command does.
The union
command has all of the options of include
. Consider this union
command:
sencha compile -classpath=sdk/src ... and \
union -namespace Ext.grid,Ext.chart and \
...
It is exactly equivalent to this pair of exclude
and include
commands:
sencha compile -classpath=sdk/src ... and \
exclude -all and \
include -namespace Ext.grid,Ext.chart and \
...
One of the most important set operations is the union of all files explicitly specified
and all of the files they require. This is the core of a build process, since this is
how you select only the set of files you need. So, if you have a small set of top-level
files to start the process, say the class MyApp.App
, you can do something like this:
sencha compile -classpath=sdk/src,app \
union -r -class MyApp.App and \
...
The union
command starts with no current set, includes only the class MyApp.App
and
then proceeds to include all the things it needs recursively. The resulting current set
is all files needed by the application.
The intersect
command is a bit less flexible in the criteria it supports: it only
accepts named sets (using -set
).
sencha compile -classpath=sdk/src,common,page1/src,page2/src \
... \
intersect -set page1,page2 and \
... \
This command above intersects the two page sets and produces their intersection as the
current set.
When dealing with more than two sets, intersect
has an option called -min
that sets
the threshold for membership in the current set. This option is discussed in more detail
in Multi-Page Ext JS Apps.
For example,
sencha compile ... \
intersect -min=2 -set page1,page2,page3 and \
...
This use of intersect
produces in the current set all files that are found in two of
the three sets specified.
In many situations, it is helpful to embed metadata in files that only the compiler will
pick up. To do this, the compiler recognizes special line comments as directives.
The list of directives is:
//@charset
//@tag
//@define
//@require
There is no standard way to specify the character encoding of a particular JS file. The
Sencha Cmd compiler, therefore, understands the following directive:
//@charset ISO-9959-1
This must be the first line of the JS file. The value to the right of charset
can be any
valid Java charset
name. The default is "UTF-8".
The charset
directive is used to describe the encoding of an input file to the compiler.
This does not effect the encoding of the output file. The content of the input file is
converted to Unicode internally.
In an ideal world, a namespace would be sufficient to define a set of interest. Sometimes,
however, a set can be quite arbitrary and even cross namespace boundaries. Rather than
move this issue to the command-line level, the compiler can track arbitrary tags in files.
Consider the example:
//@tag foo,bar
This assigns the tags foo
and bar
to the file. These tags can be used in the include
,exclude
and union
commands with their -tag
option.
In some cases, JavaScript files define classes or objects and require classes or objects
that are not expressed in terms of Ext.define
and requires
or Ext.require
. UsingExt.define
you can still say that a class requires
such things and the dynamic loader
will not complain so long as those things exist (if they do not exist, the loader will
try to load them, which will most likely fail).
To support arbitrary JavaScript approaches to defining and requiring types, the compiler
also provides these directives:
//@define Foo.bar.Thing
//@requires Bar.foo.Stuff
These directives set up the same basic metadata in the compiler that tracks what file
defines a type and what types that file requires. In most ways, then, these directives
accomplish the same thing as an Ext.define
with a requires
property.
You can use either of these directives in a file without using the other.
The compiler supports conditional compilation directives, such as the one illustrated here:
foo: function () {
//<debug>
if (sometest) {
Ext.log.warn("Something is wrong...");
}
//</debug>
this.bar();
}
This may be the most useful of the conditional directives, which you'd use for code that
you want to run in a development environment but not in production.
Important. When you use conditional compilation, remember that unless you always run
compiled code, the directives are just comments and the conditional code will be "live"
during development.
When compiling, by default, none of the preprocessor statements are examined. So in this
case, the result is development mode. If we switch on -debug
we get a very similar
result, but with the preprocessor active. In fact, the only difference is that the
preprocessor directives are removed.
For example, this command:
sencha compile -classpath=... \
-debug \
...
generates code like this:
foo: function () {
if (sometest) {
Ext.log.warn("Something is wrong...");
}
this.bar();
}
However, this command:
sencha compile -classpath=... \
-debug=false \
...
generates code like this:
foo: function () {
this.bar();
}
You can see that the if
test and the log statement are both removed.
The most general directive is if
. The if
directive tests one or more configured
options against the attributes of the directive and removes the code in the block
if any are false.
For example:
//<if debug>
//</if>
This is equivalent to the <debug>
directive.