gecko-dev/layout/doc/regression_tests.html

245 lines
18 KiB
HTML
Raw Normal View History

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN">
<!-- ***** BEGIN LICENSE BLOCK *****
- Version: MPL 1.1/GPL 2.0/LGPL 2.1
-
- The contents of this file are subject to the Mozilla Public License Version
- 1.1 (the "License"); you may not use this file except in compliance with
- the License. You may obtain a copy of the License at
- http://www.mozilla.org/MPL/
-
- Software distributed under the License is distributed on an "AS IS" basis,
- WITHOUT WARRANTY OF ANY KIND, either express or implied. See the License
- for the specific language governing rights and limitations under the
- License.
-
- The Original Code is a description how to run the layout regression tests.
-
- The Initial Developer of the Original Code is
- Bernd Mielke.
- Portions created by the Initial Developer are Copyright (C) 2___
- the Initial Developer. All Rights Reserved.
-
- Contributor(s):
-
- Alternatively, the contents of this file may be used under the terms of
- either the GNU General Public License Version 2 or later (the "GPL"), or
- the GNU Lesser General Public License Version 2.1 or later (the "LGPL"),
- in which case the provisions of the GPL or the LGPL are applicable instead
- of those above. If you wish to allow use of your version of this file only
- under the terms of either the GPL or the LGPL, and not to allow others to
- use your version of this file under the terms of the MPL, indicate your
- decision by deleting the provisions above and replace them with the notice
- and other provisions required by the LGPL or the GPL. If you do not delete
- the provisions above, a recipient may use your version of this file under
- the terms of any one of the MPL, the GPL or the LGPL.
-
- ***** END LICENSE BLOCK ***** -->
2005-01-28 20:01:46 +00:00
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<title>Layout Regression tests</title>
<style type="text/css">
body {font-family:arial}
.screen {background-color:silver; color:blue}
pre.screen {display:table-cell}
</style>
2005-01-28 20:01:46 +00:00
</head>
<body>
<h1>Table Regression Tests</h1>
<h6>Bernd Mielke</h6>
2005-01-28 20:01:46 +00:00
<h2>About This Document</h2>
<p>
This document describes my guerrilla approach to successfully running the table layout regression tests. It should encourage more people to run those tests.
2005-01-28 20:01:46 +00:00
<h2>Subject Overview</h2>
<p>
The table element and its children are one of the major layout components that can create some structure on the screen. They are widely (ab)used by page authors. Every small change in the table code can generate a bunch of bugzilla entries and a lot of seemingly unrelated dupes. With the high frequency of checkins in to the source tree it is difficult and time consuming to track the regression down to a single checkin. The risk of becoming very infamous can be reduced by running the layout regression tests.
2005-01-28 20:01:46 +00:00
<h2>Prerequisites</h2>
<p>
In order to run these tests, you will need to have:
<ol>
<li>a debug build with a working viewer, an optimized build will not work!
<li>a mozilla tree with the testfiles at <code>%MOZ_SRC%/layout/html/tests</code>,
<li> <code>DISABLE_TESTS</code> should not be defined,
<li>patience and time.
</ol>
2005-01-28 20:01:46 +00:00
<h2>Test Runs</h2>
<ul>
<li>Go to the <code>%MOZ_SRC%/layout/html/tests/block</code> directory.
<li> Avoid the nasty assertions to pop up by <code>set XPCOM_DEBUG_BREAK=warn</code>, the world would be a much better place if we could <a href="http://bugzilla.mozilla.org/show_bug.cgi?id=152015">first fix those assertions</a>.
<li>The following shell scripts need a <code>sh</code> before the command under WinXX.
<li>Execute <code>rtest.sh baseline</code>
<li>Execute <code>rtest.sh verify&gt;outputfile.txt</code>
<li>Make a note which tests have failed (these are false positives. If you are in bad mood file a bug against Chris Waterson about them)
<li>you can grep the failures by:<br>
<code>grep 'rgd failed' &lt;outputfile.txt &gt;false_positive.txt</code>
<li> Make a copy of <code>gklayout.dll</code> and <code>gkcontent.dll</code> in the <code>dist</code> directory. If you would like to switch back, you will need them.
<li>Make your changes to the source, compile and install.
<li>execute in the <code>layout/html/tests/block</code> directory <code>rtest.sh verify &gt;regression.txt</code>
<li>Check how many tests have failed and analyse your results:
<ul>
<li>If the regression tests have failed at the same place as the previous diagnosed false positives =&gt; ignore
<ol>
<li>copy <code>rtest.sh</code> to <code>my_rtest.sh</code>
<li>create a new directory <code>my_files</code>.
<li>create in this directory a file <code>file_list.txt</code>, where you put the url's of your remaining test failures
<li>change <code>my_rtest.sh</code> to use only directory <code>my_files</code>
<li>go back to the original gklayout.dll and gkcontent.dll and repeat the regression test procedure on a much smaller set now.
</ol>
<li>Try to figure out what the other regressions are: Are they
improving the picture or are they regressions?
<li>Note the difficult to explain testfailures.
</ul>
<li> Change the source if necessary until the regression tests indicate that your patch does not hork the block and table layout
<li> Submit your patch and lists the remaining differences in the bug
</ul>
<p>Be prepared to cycle.
<h2>How the layout regression tests work</h2>
<p>First the URL's to test are red from <code>file_list.txt</code>.<p>
<table><caption>snippet from file_list.txt</caption><tr><td>
<pre>file:///s|/mozilla/layout/html/tests/table/dom/appendCells1.html
file:///s|/mozilla/layout/html/tests/table/dom/appendCellsRebuild1.html
file:///s|/mozilla/layout/html/tests/table/dom/appendCol1.html
file:///s|/mozilla/layout/html/tests/table/dom/appendCol2.html
file:///s|/mozilla/layout/html/tests/table/dom/appendColGroup1.html
file:///s|/mozilla/layout/html/tests/table/dom/appendRowsExpand1.html</pre>
</td></tr></table>
<p>The regression tests compare the frame tree dumps recorded at two different times. The URL's to test are red from <code>file_list.txt</code>
For every file in <code>file_list.txt</code>viewer takes the root frame and dumps the whole tree via
<a href="http://lxr.mozilla.org/seamonkey/search?string=nsFrame%3A%3ADumpRegression">
nsFrame::DumpRegression</a>.
<p>
The frame tree dumps are invoked by the <code>-o</code> switch to the <code>viewer</code> (see <a href="http://lxr.mozilla.org/seamonkey/ident?i=PrintHelpInfo">commandline arguments </a>).
When <code>-o</code> is specified together with a output directory for instance (<code>-o s:\mozilla\layout\html\tests\table\dom\</code>),
the frame trees are dumped to separate files in that directory for all URL's in <code>file_list.txt</code> (see:
<a href="http://lxr.mozilla.org/seamonkey/search?string=nsWebCrawler%3A%3ADumpRegression">
nsWebCrawler::DumpRegressionData</a>) .
<p>Their name is generated by replacing the old extension <code>.html</code> or <code>.xml</code> with <code>.rgd</code>. For <code>file:///s|/mozilla/layout/html/tests/table/dom/appendCells1.html</code> a file <code>appendCells1.rgd</code> will be created in the <code>s:\mozilla\layout\html\tests\table\dom\</code> directory.
<p>
There is a special type of frame dumps - the printing regression tests, they are invoked by the <code>-Prt</code> command line argument to the <code>viewer</code> and include first a display then a frame dump taking into account the printer pages. A typical beginning of a dump (<code>*.rgd</code> file) looks like:<p>
<pre class="screen">
&lt;frame va="15022440" type="Viewport(-1)" state="270340" parent="0"&gt;
&lt;view va="47171904"&gt;
&lt;/view&gt;
&lt;stylecontext va="15022232"&gt;
&lt;font serif 240 240 0 /&gt;
&lt;color data="-16777216"/&gt;
&lt;background data="0 2 3 -1 0 0 "/&gt;
&lt;spacing data="left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null left: 1[0x1]enum top: 1[0x1]enum right: 1[0x1]enum bottom: 1[0x1]enum left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null 1[0x1]enum 0" /&gt;
&lt;list data="100 100 " /&gt;
&lt;position data="left: Auto top: Auto right: Auto bottom: Auto Auto 0[0x0]tw Null Auto 0[0x0]tw Null 0 Auto " /&gt;
&lt;text data="0 0 0 Normal Normal 0[0x0]tw Normal " /&gt;
&lt;textreset data="0 10[0xa]enum " /&gt;
&lt;display data="0 1 0 0 0 0 0 0 0 0 0 0 " /&gt;
&lt;visibility data="0 1 1.000000" /&gt;
&lt;table data="0 0 4 -1 1 " /&gt;
&lt;tableborder data="1 Null Null 0 2 " /&gt;
&lt;content data="0 0 0 Null " /&gt;
&lt;quotes data="0 " /&gt;
&lt;ui data="3 0 0 1 " /&gt;
&lt;uireset data="7 0 4" /&gt;
&lt;xul data="0 0 0 0 0 1 &lt;svg data="0 1.000000 1.000000 0 1.000000" /&gt;
&lt;/stylecontext&gt;</pre>
<p>The first run of the regression tests is started with the argument <code>baseline</code> and it creates the reference frame dumps.
<p>The <code>baseline</code> log will look like:
<pre class="screen">
Type Manifest File: C:\MOZ_SOUR\MOZILLA\MOZILLA\DIST\WIN32_D.OBJ\BIN\components\xpti.dat
nsNativeComponentLoader: autoregistering begins.
nsNativeComponentLoader: autoregistering succeeded
Going to create the event queue
WEBSHELL+ = 1
Note: verifyreflow is disabled
Note: styleverifytree is disabled
Note: frameverifytree is disabled
+++ file:///s|/mozilla/layout/html/tests/table/dom/appendCells1.html: done loading (11810 msec)
appendCells1.rgd - being written
+++ file:///s|/mozilla/layout/html/tests/table/dom/appendCellsRebuild1.html: done loading (1260 msec)
appendCellsRebuild1.rgd - being written
+++ file:///s|/mozilla/layout/html/tests/table/dom/appendCol1.html: done loading (270 msec)
appendCol1.rgd - being written
+++ file:///s|/mozilla/layout/html/tests/table/dom/appendCol2.html: done loading (220 msec)
appendCol2.rgd - being written
+++ file:///s|/mozilla/layout/html/tests/table/dom/appendColGroup1.html: done loading (390 msec)
appendColGroup1.rgd - being written
+++ file:///s|/mozilla/layout/html/tests/table/dom/appendRowsExpand1.html: done loading (220 msec)
appendRowsExpand1.rgd - being written
</pre>
<p>
The second run with the argument <code>verify</code> creates first the frame dumps in a subdirectory (<code>verify</code>) and <a href="http://lxr.mozilla.org/seamonkey/search?string=nsFrameUtil%3A%3ACompareTrees">compares</a> each frame dump with the corresponding reference frame dump. If they differ in critical points the test fails.
<p>A typical part of the <code>verify</code> log would look like:
<pre class="screen">
Type Manifest File: C:\MOZ_SOUR\MOZILLA\MOZILLA\DIST\WIN32_D.OBJ\BIN\components\xpti.dat
nsNativeComponentLoader: autoregistering begins.
nsNativeComponentLoader: autoregistering succeeded
Going to create the event queue
WEBSHELL+ = 1
Note: verifyreflow is disabled
Note: styleverifytree is disabled
Note: frameverifytree is disabled
+++ file:///s|/mozilla/layout/html/tests/table/dom/appendCells1.html: done loading (2200 msec)
frame bbox mismatch: 0,285,930,435 vs. 0,285,1305,435
Node 1:
TableOuter(table)(1) 0x10004 0,285,930,435, |null attr|-16777216|left: 0[0x0]tw top: 0[0x0]tw right: 0[0x0]tw bottom: 0[0x0]tw left: 0[0x0]tw top: 0[0x0]tw right: 0[0x0]tw bottom: 0[0x0]tw left: 1[0x1]enum top: 1[0x1]enum right: 1[0x1]enum bottom: 1[0x1]enum left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null 1[0x1]enum 0|100 100 |left: Inherit top: Inherit right: Inherit bottom: Inherit 0[0x0]tw 0[0x0]tw Null 0[0x0]tw 0[0x0]tw Null 0 Auto |0 0 0 Normal Normal 0[0x0]tw Normal |0 8 0 0 0 0 0 1 0 0 0 0 |0 0 4 -1 1 |0 0 0 Null
Node 2:
TableOuter(table)(1) 0x10004 0,285,1305,435, |null attr|-16777216|left: 0[0x0]tw top: 0[0x0]tw right: 0[0x0]tw bottom: 0[0x0]tw left: 0[0x0]tw top: 0[0x0]tw right: 0[0x0]tw bottom: 0[0x0]tw left: 1[0x1]enum top: 1[0x1]enum right: 1[0x1]enum bottom: 1[0x1]enum left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null 1[0x1]enum 0|100 100 |left: Inherit top: Inherit right: Inherit bottom: Inherit 0[0x0]tw 0[0x0]tw Null 0[0x0]tw 0[0x0]tw Null 0 Auto |0 0 0 Normal Normal 0[0x0]tw Normal |0 8 0 0 0 0 0 1 0 0 0 0 |0 0 4 -1 1 |0 0 0 Null
frame bbox mismatch: 0,0,930,435 vs. 0,0,1305,435
Node 1:
Table(table)(1) 0x10004 0,0,930,435, |null attr|-16777216|left: Null top: 0[0x0]tw right: Null bottom: 0[0x0]tw left: Null top: Null right: Null bottom: Null left: 15[0xf]tw top: 15[0xf]tw right: 15[0xf]tw bottom: 15[0xf]tw left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null 1[0x1]enum 0|100 100 |left: Auto top: Auto right: Auto bottom: Auto Auto 0[0x0]tw Null Auto 0[0x0]tw Null 2 Auto |0 0 0 Normal Normal 0[0x0]tw Normal |0 8 0 0 0 0 0 0 0 0 0 0 |0 8 4 -1 1 |0 0 0 Null
Node 2:
Table(table)(1) 0x10004 0,0,1305,435, |null attr|-16777216|left: Null top: 0[0x0]tw right: Null bottom: 0[0x0]tw left: Null top: Null right: Null bottom: Null left: 15[0xf]tw top: 15[0xf]tw right: 15[0xf]tw bottom: 15[0xf]tw left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null 1[0x1]enum 0|100 100 |left: Auto top: Auto right: Auto bottom: Auto Auto 0[0x0]tw Null Auto 0[0x0]tw Null 2 Auto |0 0 0 Normal Normal 0[0x0]tw Normal |0 8 0 0 0 0 0 0 0 0 0 0 |0 8 4 -1 1 |0 0 0 Null
frame bbox mismatch: 45,15,840,405 vs. 45,15,1215,405
Node 1:
TableColGroup(table)(1) 0x80010004 45,15,840,405, |null attr|-16777216|left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null left: 1[0x1]enum top: 1[0x1]enum right: 1[0x1]enum bottom: 1[0x1]enum left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null 1[0x1]enum 0|100 100 |left: Auto top: Auto right: Auto bottom: Auto Auto 0[0x0]tw Null Auto 0[0x0]tw Null 0 Auto |0 0 0 Normal Normal 0[0x0]tw Normal |0 12 0 0 0 0 0 0 0 0 0 0 |0 0 4 -1 1 |0 0 0 Null
Node 2:
TableColGroup(table)(1) 0x80010004 45,15,1215,405, |null attr|-16777216|left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null left: 1[0x1]enum top: 1[0x1]enum right: 1[0x1]enum bottom: 1[0x1]enum left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null 1[0x1]enum 0|100 100 |left: Auto top: Auto right: Auto bottom: Auto Auto 0[0x0]tw Null Auto 0[0x0]tw Null 0 Auto |0 0 0 Normal Normal 0[0x0]tw Normal |0 12 0 0 0 0 0 0 0 0 0 0 |0 0 4 -1 1 |0 0 0 Null
frame state mismatch: 0x30010406 vs. 0x30010004
Node 1:
TableCol(table)(1) 0x30010406 0,0,0,0, |null attr|-16777216|left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null left: 1[0x1]enum top: 1[0x1]enum right: 1[0x1]enum bottom: 1[0x1]enum left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null 1[0x1]enum 0|100 100 |left: Auto top: Auto right: Auto bottom: Auto Auto 0[0x0]tw Null Auto 0[0x0]tw Null 0 Auto |0 0 0 Normal Normal 0[0x0]tw Normal |0 11 0 0 0 0 0 0 0 0 0 0 |0 0 4 -1 1 |0 0 0 Null
Node 2:
TableCol(table)(1) 0x30010004 870,0,345,405, |null attr|-16777216|left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null left: 1[0x1]enum top: 1[0x1]enum right: 1[0x1]enum bottom: 1[0x1]enum left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null 1[0x1]enum 0|100 100 |left: Auto top: Auto right: Auto bottom: Auto Auto 0[0x0]tw Null Auto 0[0x0]tw Null 0 Auto |0 0 0 Normal Normal 0[0x0]tw Normal |0 11 0 0 0 0 0 0 0 0 0 0 |0 0 4 -1 1 |0 0 0 Null
frame bbox mismatch: 0,0,0,0 vs. 870,0,345,405
Node 1:
TableCol(table)(1) 0x30010406 0,0,0,0, |null attr|-16777216|left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null left: 1[0x1]enum top: 1[0x1]enum right: 1[0x1]enum bottom: 1[0x1]enum left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null 1[0x1]enum 0|100 100 |left: Auto top: Auto right: Auto bottom: Auto Auto 0[0x0]tw Null Auto 0[0x0]tw Null 0 Auto |0 0 0 Normal Normal 0[0x0]tw Normal |0 11 0 0 0 0 0 0 0 0 0 0 |0 0 4 -1 1 |0 0 0 Null
Node 2:
TableCol(table)(1) 0x30010004 870,0,345,405, |null attr|-16777216|left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null left: 1[0x1]enum top: 1[0x1]enum right: 1[0x1]enum bottom: 1[0x1]enum left: Null top: Null right: Null bottom: Null left: Null top: Null right: Null bottom: Null 1[0x1]enum 0|100 100 |left: Auto top: Auto right: Auto bottom: Auto Auto 0[0x0]tw Null Auto 0[0x0]tw Null 0 Auto |0 0 0 Normal Normal 0[0x0]tw Normal |0 11 0 0 0 0 0 0 0 0 0 0 |0 0 4 -1 1 |0 0 0 Null
regression test s:\mozilla\layout\html\tests\table\dom\verify\appendCells1.rgd failed
+++ file:///s|/mozilla/layout/html/tests/table/dom/appendCellsRebuild1.html: done loading (390 msec)
044regression test s:\mozilla\layout\html\tests\table\dom\verify\appendCellsRebuild1.rgd passed
+++ file:///s|/mozilla/layout/html/tests/table/dom/appendCol1.html: done loading (280 msec)
regression test s:\mozilla\layout\html\tests\table\dom\verify\appendCol1.rgd passed
+++ file:///s|/mozilla/layout/html/tests/table/dom/appendCol2.html: done loading (220 msec)
regression test s:\mozilla\layout\html\tests\table\dom\verify\appendCol2.rgd passed
+++ file:///s|/mozilla/layout/html/tests/table/dom/appendColGroup1.html: done loading (220 msec)
regression test s:\mozilla\layout\html\tests\table\dom\verify\appendColGroup1.rgd passed
</pre>
<p>
The first regression test has failed and it is marked so.
2005-01-28 20:01:46 +00:00
<h2>Adding new regression tests</h2>
<p>
Once you have checked in the code:
<ul>
<li> please add your testcase for the bug to the regression tests.
<li>A testcase should only contain local files.
<li>Place the images in <code>table/images</code> (don't forget <code> cvs commit -kb </code>)
<li>and the testfile in the <code>table/bugs</code> directory.
<li>Update the <code>file_list<b>x</b>.txt</code> in order to include your file.
</ul>
<p>I would like to thank Chris Karnaze for his guidance and fantasai for his language support.
2005-01-28 20:01:46 +00:00
<div style="text-align:right"><i>&lt;regression.html&gt;&nbsp; &lt;Last updated:
2002-06-16 <a href="mailto:bernd.mielke@snafu.de">Bernd Mielke</a></i></div>
</body>
</html>