gecko-dev/js/tests/README-jsDriver.html
2001-07-27 21:24:49 +00:00

345 lines
14 KiB
HTML

<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML//EN">
<html>
<head>
<title>jsDriver.pl</title>
</head>
<body bgcolor="white">
<h1 align="right">jsDriver.pl</h1>
<dl>
<dt><b>NAME</b></dt>
<dd>
<b>jsDriver.pl</b> - execute JavaScript programs in various shells in
batch or single mode, reporting on failures encountered.
<br>
<br>
<dt><b>SYNOPSIS</b></dt>
<dd>
<table>
<tr>
<td align="right" valign="top">
<code>
<b>jsDriver.pl</b>
</code>
</td>
<td>
<code>
[-hkt] [-b BUGURL] [-c CLASSPATH] [-f OUTFILE]
[-j JAVAPATH] [-l TESTLIST ...] [-L NEGLIST ...] [-p TESTPATH]
[-s SHELLPATH] [-u LXRURL] [--help] [--confail] [--trace]
[--classpath=CLASSPATH] [--file=OUTFILE] [--javapath=JAVAPATH]
[--list=TESTLIST] [--neglist=TESTLIST] [--testpath=TESTPATH]
[--shellpath=SHELLPATH] [--lxrurl=LXRURL] {-e ENGINETYPE |
--engine=ENGINETYPE}
</code>
</td>
</tr>
</table>
<br>
<br>
<dt><b>DESCRIPTION</b></dt>
<dd>
<b>jsDriver.pl</b> is normally used to run a series of tests against
one of the JavaScript shells. These tests are expected to be laid out
in a directory structure exactly three levels deep. The first level
is considered the <b>root</b> of the tests, subdirectories under the
<b>root</b> represent <b>Test Suites</b> and generally mark broad
categories such as <i>ECMA Level 1</i> or <i>Live Connect 3</i>. Under the
<b>Test Suites</b> are the <b>Test Categories</b>, which divide the
<b>Test Suite</b> into smaller categories, such as <i>Execution Contexts</i>
or <i>Lexical Rules</i>. Testcases are located under the
<B>Test Categories</b> as normal JavaScript (*.js) files.
<p>
If a file named <b>shell.js</b> exists in either the
<b>Test Suite</b> or the <b>Test Category</b> directory, it is
loaded into the shell before the testcase. If <b>shell.js</b>
exists in both directories, the version in the <b>Test Suite</b>
directory is loaded <i>first</i>, giving the version associated with
the <b>Test Category</b> the ability to override functions previously
declared. You can use this to
create functions and variables common to an entire suite or category.
<p>
Testcases can report failures back to <b>jsDriver.pl</b> in one of
two ways. The most common is to write a line of text containing
the word <code>FAILED!</code> to <b>STDOUT</b> or <b>STDERR</b>.
When the engine encounters a matching line, the test is marked as
failed, and any line containing <code>FAILED!</code> is displayed in
the failure report. The second way a test case can report failure is
to return an unexpected exit code. By default, <b>jsDriver.pl</b>
expects all test cases to return exit code 0, although a test
can output a line containing <code>EXPECT EXIT <i>n</i></code> where
<i>n</i> is the exit code the driver should expect to see. Testcases
can return a nonzero exit code by calling the shell function
<code>quit(<i>n</i>)</code> where <code><i>n</i></code> is the
code to exit with. The various JavaScript shells report
non-zero exit codes under the following conditions:
<center>
<table border="1">
<tr>
<th>Reason</th>
<th>Exit Code</th>
</tr>
<tr>
<td>
Engine initialization failure.
</td>
<td>
1
</td>
</tr>
<tr>
<td>
Invalid argument on command line.
</td>
<td>
2
</td>
</tr>
<tr>
<td>
Runtime error (uncaught exception) encountered.
</td>
<td>
3
</td>
</tr>
<tr>
<td>
File argument specified on command line not found.
</td>
<td>
4
</td>
</tr>
<tr>
<td>
Reserved for future use.
</td>
<td>
5-9
</td>
</tr>
</table>
</center>
<br>
<br>
<dt><b>OPTIONS</b></dt>
<dd>
<dl>
<dt><b>-b URL, --bugurl=URL</b></dt>
<dd>
Bugzilla URL. When a testcase writes a line in the format
<code>BUGNUMBER <i>n</i></code> to <b>STDOUT</b> or <b>STDERR</b>,
<b>jsDriver.pl</b> interprets <code><i>n</i></code> as a bugnumber
in the <a href="http://bugzilla.mozilla.org">BugZilla</a> bug
tracking system. In the event that a testcase which has specified
a bugnumber fails, a hyperlink to the BugZilla database
will be included in the output by prefixing the bugnumber with the
URL specified here. By default, URL is assumed to be
&quot;http://bugzilla.mozilla.org/show_bug.cgi?id=&quot;.
<br>
<br>
<a name="classpath"></a>
<dt><b>-c PATH, --classpath=PATH</b></dt>
<dd>
Classpath to pass the the Java Virtual Machine. When running tests
against the <b>Rhino</b> engine, PATH will be passed in as the value
to an argument named &quot;-classpath&quot;. If your particular JVM
does not support this option, it is recommended you specify your
class path via an environment setting. Refer to your JVM
documentation for more details about CLASSPATH.
<br>
<br>
<dt><b>-e TYPE ..., --engine=TYPE ...</b></dt>
<dd>
Required. Type of engine(s) to run the tests against. TYPE can be
one or more of the following values:
<center>
<table border="1">
<tr>
<th>TYPE</th>
<th>Engine</th>
</tr>
<tr>
<td>lcopt</td>
<td>LiveConnect, optimized</td>
</tr>
<tr>
<td>lcdebug</td>
<td>LiveConnect, debug</td>
</tr>
<tr>
<td>rhino</td>
<td>Rhino compiled mode</td>
</tr>
<tr>
<td>rhinoi</td>
<td>Rhino interpreted mode</td>
</tr>
<tr>
<td>rhinoms</td>
<td>Rhino compiled mode for the Microsoft VM (jview)</td>
</tr>
<tr>
<td>rhinomsi</td>
<td>Rhino interpreted mode for the Microsoft VM (jview)</td>
</tr>
<tr>
<td>smopt</td>
<td>Spider-Monkey, optimized</td>
</tr>
<tr>
<td>smdebug</td>
<td>Spider-Monkey, debug</td>
</tr>
<tr>
<td>xpcshell</td>
<td>XPConnect shell</td>
</tr>
</table>
</center>
<br>
<br>
<dt><b>-f FILE, --file=FILE</b></dt>
<dd>
Generate html output to the HTML file named by FILE. By default,
a filename will be generated using a combination of the engine type
and a date/time stamp, in the format:
<code>results-<i>&lt;engine-type&gt;</i>-<i>&lt;date-stamp&gt;</i>.html</code>
<br>
<br>
<dt><b>-h, --help</b></dt>
<dd>
Prints usage information.
<br>
<br>
<dt><b>-j PATH, --javapath=PATH</b></dt>
<dd>
Set the location of the Java Virtual Machine to use when running
tests against the <b>Rhino</b> engine. This can be used to test
against multiple JVMs on the same system.
<br>
<br>
<dt><b>-k, --confail</b></dt>
<dd>
Log failures to the console. This will show any failures, as they
occur, on <b>STDERR</b> in addition to creating the HTML results
file. This can be useful for times when it may be
counter-productive to load an HTML version of the results each time
a test is re-run.
<br>
<br>
<dt><b>-l FILE ..., --list=FILE ...</b></dt>
<dd>
Specify a list of tests to execute. FILE can be a plain text file
containing a list of testcases to execute, a subdirectory
in which to
<a href="http://www.instantweb.com/~foldoc/foldoc.cgi?query=grovel">grovel</a>
for tests, or a single testcase to execute. Any number of FILE
specifiers may follow this option. The driver uses the fact that a
valid testcase should be a file ending in .js to make the distinction
between a file containing a list of tests and an actual testcase.
<br>
<br>
<dt><b>-L FILE ..., --neglist=FILE ...</b></dt>
<dd>
Specify a list of tests to skip. FILE has the same meaning as in
the <b>-l</b> option. This option is evaluated after
<b>all</b> <b>-l</b> and <b>--list</b> options, allowing a user
to subtract a single testcase, a directory of testcases, or a
collection of unrelated testcases from the execution list.
<br>
<br>
<dt><b>-p PATH, --testpath=PATH</b></dt>
<dd>
Directory holding the &quot;Test Suite&quot; subdirectories. By
default this is ./
<br>
<br>
<dt><b>-s PATH, --shellpath=PATH</b></dt>
<dd>
Directory holding the JavaScript shell. This can be used to override
the automatic shell location <b>jsDriver.pl</b> performs based on
you OS and engine type. For Non <b>Rhino</b> engines, this
includes the name of the executable as well as the path. In
<b>Rhino</b>, this path will be appended to your
<a href="#classpath">CLASSPATH</a>. For the
<b>SpiderMonkey</b> shells, this value defaults to
../src/&lt;Platform-and-buildtype-specific-directory&gt;/[js|jsshell],
for the
<b>LiveConnect</b> shells,
../src/liveconnect/src/&lt;Platform-and-buildtype-specific-directory&gt;/lschell
and for the <b>xpcshell</b> the default is the value of your
<code>MOZILLA_FIVE_HOME</code> environment variable. There is no
default (as it is usually not needed) for the <b>Rhino</b> shell.
<br>
<br>
<dt><b>-t, --trace</b></dt>
<dd>
Trace execution of <b>jsDriver.pl</b>. This option is primarily
used for debugging of the script itself, but if you are interested in
seeing the actual command being run, or generally like gobs of
useless information, you may find it entertaining.
<br>
<br>
<dt><b>-u URL, --lxrurl=URL</b></dt>
<dd>
Failures listed in the HTML results will be hyperlinked to the
lxr source available online by prefixing the test path and
name with this URL. By default, URL is
http://lxr.mozilla.org/mozilla/source/js/tests/
<br>
<br>
</dl>
<dt><b>SEE ALSO</b></dt>
<dd>
<a href="http://lxr.mozilla.org/mozilla/source/js/tests/jsDriver.pl">jsDriver.pl</a>,
<a href="http://lxr.mozilla.org/mozilla/source/js/tests/mklistpage.pl">mklistpage.pl</a>,
<a href="http://www.mozilla.org/js/">http://www.mozilla.org/js/</a>,
<a href="http://www.mozilla.org/js/tests/library.html">http://www.mozilla.org/js/tests/library.html</a>
<br>
<br>
<dt><b>REQUIREMENTS</b></dt>
<dd>
<b>jsDriver.pl</b> requires the
<a href="http://search.cpan.org/search?module=Getopt::Mixed">Getopt::Mixed</a>
perl package, available from <a href="http://www.cpan.org">cpan.org</a>.
<br>
<br>
<dt><b>EXAMPLES</b></dt>
<dd>
<code>perl jsDriver.pl -e smdebug -L lc*</code><br>
Executes all tests EXCEPT the liveconnect tests against the
SpiderMonkey debug shell, writing the results
to the default result file. (NOTE: Unix shells take care of wildcard
expansion, turning <code>lc*</code> into <code>lc2 lc3</code>. Under
a DOS shell, you must explicitly list the directories.)
<p>
<code>perl jsDriver.pl -e rhino -L rhino-n.tests</code><br>
Executes all tests EXCEPT those listed in the
<code>rhino-n.tests</code> file.
<p>
<code>perl -I/home/rginda/perl/lib/ jsDriver.pl -e lcopt -l lc2
lc3 -f lcresults.html -k</code><br>
Executes ONLY the tests under the <code>lc2</code> and <code>lc3</code>
directories against the LiveConnect shell. Results will be written to
the file <code>lcresults.html</code> <b>AND</b> the console. The
<code>-I</code> option tells perl to look for modules in the
<code>/home/rginda/perl/lib</code> directory (in addition to the
usual places), useful if you do not have root access to install new
modules on the system.
</dl>
<hr>
Author: Robert Ginda<br>
Currently maintained by <i><a href="mailto:pschwartau@netscape.com">Phil Schwartau</a> </i><br>
<!-- Created: Thu Dec 2 19:08:05 PST 1999 -->
</body>
</html>