doc changes, not affecting the build.

This commit is contained in:
karnaze%netscape.com 2002-08-04 06:05:25 +00:00
Родитель 8008716d17
Коммит 8f2a63ae19
1 изменённых файлов: 222 добавлений и 225 удалений

Просмотреть файл

@ -77,26 +77,65 @@
background: #00f;
}
table.tree {
table {
margin: auto;
}
table.tree td {
text-align: center;
table.box {
border-collapse: collapse;
border: 2px solid black;
-moz-box-sizing: border-box;
width:98%;
height:96%;
}
row {
border: 2px solid black;
}
row.title {
border: hidden;
}
table td {
text-align: left;
empty-cells: hide;
}
table td.box {
height:100%;
border: 2px solid black;
}
table td.title {
border-top: hidden;
border-right: hidden;
border-left: hidden;
font-size: .5em;
}
table td.title2 {
border-bottom: hidden;
border-right: hidden;
border-left: hidden;
font-size: .5em;
}
table.tree tr:not(.arrows) td {
font-family: monospace;
border: 0.1em solid;
}
span.html {
color:purple;
}
</style>
</head>
<body>
<div class="slide" id="s0">
<div class="nav">
<a href="#s25">&lt;</a>
<a href="#s1">&gt;</a>
</div>
<table style="margin: auto"><tr><td>
@ -119,9 +158,8 @@
<li>Review of reflow</li>
<li>Table frames</li>
<li>Table reflow</li>
<li>Intro to paganated reflow</li>
<li>Table paganated reflow</li>
<li>Areas not covered</li>
<li>Intro to paginated reflow</li>
<li>Table paginated reflow</li>
</ul>
</div>
@ -263,7 +301,7 @@
on the table, row groups, rows, cells the preferred and max element sizes of the cells
(from the pass 1 reflow), and considers colspans</li>
<li>Pass 2 - cell widths are constrained by the column widths (heights are only
constrained in paganated mode).</li>
constrained in paginated mode).</li>
<li>The row group figures out the row heights given the its style height constraints
its rows and cells and the actual heights of its rows and cells from the pass 2 reflow</li>
<li>If the table has a style height, it allocates extra height to its row groups, rows and cells.</li>
@ -278,7 +316,65 @@
<a href="#s8">&lt;</a>
<a href="#s10">&gt;</a>
</div>
<h1>Table Reflow</h1>
<h1>Table Reflow Example</h1>
<table width=100%>
<tr>
<td width=100%>
<pre style="font-size:.6em;">
tblO 030176CC r=0 a=8940,UC c=0,0 cnt=429
tbl 030178C4 r=0 a=8940,UC c=4470,UC cnt=430
rowG 03017A7C r=0 a=UC,UC c=UC,UC cnt=431
row 03017C08 r=0 a=UC,UC c=UC,UC cnt=432
cell 03017DA8 r=0 a=UC,UC c=UC,UC cnt=433
block 03017E08 r=0 a=UC,UC c=UC,UC cnt=434
block 03017E08 d=870,300 me=480
cell 03017DA8 d=930,360 me=540
cell 0301A8CC r=0 a=UC,UC c=UC,UC cnt=436
block 0301A92C r=0 a=UC,UC c=UC,UC cnt=437
block 0301A92C d=1335,300 me=465
cell 0301A8CC d=1395,360 me=525
row 03017C08 d=UC,360
rowG 03017A7C d=UC,360
rowG 03017A7C r=2 a=4470,UC c=4470,UC cnt=442
row 03017C08 r=2 a=4470,UC c=4470,UC cnt=443
cell 03017DA8 r=2 a=1755,UC c=1695,UC cnt=444
block 03017E08 r=2 a=1695,UC c=1695,UC cnt=445
block 03017E08 d=1695,300
cell 03017DA8 d=1755,360
cell 0301A8CC r=2 a=2625,UC c=2565,UC cnt=446
block 0301A92C r=2 a=2565,UC c=2565,UC cnt=447
block 0301A92C d=2565,300
cell 0301A8CC d=2625,360
row 03017C08 d=4470,360
rowG 03017A7C d=4470,360
tbl 030178C4 d=4500,450
tblO 030176CC d=4500,450
</td>
<td>
<pre style="font-size:.7em;">
&lt;<span class=html>table</span> width=300&gt;
&lt;<span class=html>tr</span>&gt;
&lt;<span class=html>td</span>&gt;foo&lt;<span class=html>/td</span>&gt;
&lt;<span class=html>td</span>&gt;bar zap&lt;<span class=html>/td</span>&gt;
&lt;<span class=html>/tr</span>&gt;
&lt;<span class=html>/table</span>&gt;
</pre>
<BR>
<BR>
<pre style="font-size:.7em;">
Key:
r = reflow reason,
0 (initial),
2 (resize)
a = avail w, h
c = computed w, h
d = desired w, h
me = max elem w
<pre>
</td>
</tr>
</table>
</div>
<div class="slide" id="s10">
@ -341,7 +437,7 @@
<li>When there is a % height frame inside a cell without a computed height</li>
<ul class=nested>
<li>the frame will never get a chance to size based on the final cell height</li>
<li>in paganated mode when there is a height on the table, the table doesn't allocate
<li>in paginated mode when there is a height on the table, the table doesn't allocate
extra height to rows until after it does a pass 2 reflow and then it is too late</li>
</ul>
</li>
@ -355,7 +451,43 @@
<a href="#s13">&lt;</a>
<a href="#s15">&gt;</a>
</div>
<h1>Special height reflow</h1>
<h1>Special Reflow Example</h1>
<table width=100%>
<tr>
<td>
<pre style="font-size:.7em;">
&lt;<span class=html>table</span> border=2 width=300&gt;
&lt;<span class=html>tr</span>&gt;
&lt;<span class=html>td</span>&gt;
This cell's width is
constrained by the
table and image widths.
Its exact height is
hard to determine.
&lt;<span class=html>/td</span>&gt;
&lt;<span class=html>td</span>&gt;
&lt;<span class=html>img</span> src=raptor.jpg
width=200 height=100%>
&lt;<span class=html>/td</span>&gt;
&lt;<span class=html>/tr</span>&gt;
&lt;<span class=html>/table</span>&gt;
</pre>
</td>
<td>
<table border=2 width=300>
<tr>
<td style="font-size:.7em;">This cell's width is constrained
by the table and image widths. Its height
is hard to determine.
<td>
<img src="raptor.jpg"
width=200 height=100%>
</td>
</tr>
</table>
</tr>
</table>
</div>
<div class="slide" id="s15">
@ -412,7 +544,7 @@
<a href="#s17">&lt;</a>
<a href="#s19">&gt;</a>
</div>
<h1>Intro to paganated reflow</h1>
<h1>Intro to paginated reflow</h1>
<ul>
<li>When a reflowee's available height is constrained, it may not fit and need to either
split/continue on the next page, or start on the next page.</li>
@ -431,7 +563,81 @@
<a href="#s18">&lt;</a>
<a href="#s20">&gt;</a>
</div>
<h1>Intro to paganated reflow</h1>
<h1>Pagination Illustration</h1>
<table class=box style="height:500px">
<tr><td class="title">nsSimplePageSequence</td></tr>
<tr>
<td class=box>
<table class=box style="height:40%;">
<tr><td class=title>nsPageFrame</td></tr>
<tr>
<td class=box>
<table class=box style="height:100%;">
<tr><td class=title>nsPageContentFrame</td></tr>
<tr>
<td class=box style="border-bottom: hidden">
<table class=box style="height:100%; border-bottom:hidden">
<tr><td class=title>areaFrame (html)</td></tr>
<tr>
<td class=box>
<table class=box style="height:100%; border-bottom:hidden">
<tr><td class=title>blockFrame (body)</td></tr>
<tr>
<td class=box>
<table class=box style="height:100%; border-bottom:hidden">
<tr><td class=title>nsTableOuterFrame</td></tr>
<tr>
<td class=box></td>
</tr>
</table>
</td>
</tr>
</table>
</td>
</tr>
</table>
</tr>
</table>
</td>
</tr>
</table>
<BR>
<table class=box style="height:40%;">
<tr>
<td class=box>
<table class=box style="height:100%;">
<tr>
<td class=box style="border-top: hidden">
<table class=box style="height:100%; border-top:hidden">
<tr>
<td class=box>
<table class=box style="height:100%; border-top:hidden">
<tr>
<td class=box>
<table class=box style="height:100%; border-top:hidden">
<tr>
<td class=box></td>
</tr>
<tr><td class=title2>nsTableOuterFrame</td></tr>
</table>
</td>
</tr>
<tr><td class=title2>blockFrame (body)</td></tr>
</table>
</td>
</tr>
<tr><td class=title2>areaFrame (html)</td></tr>
</table>
</tr>
<tr><td class=title2>nsPageContentFrame continued</td></tr>
</table>
</td>
</tr>
<tr><td class=title2>nsPageFrame continued</td></tr>
</table>
</td>
</tr>
</table>
</div>
<div class="slide" id="s20">
@ -439,7 +645,7 @@
<a href="#s19">&lt;</a>
<a href="#s21">&gt;</a>
</div>
<h1>Intro to paganated reflow</h1>
<h1>Intro to paginated reflow</h1>
<ul>
<li>The page sequence starts with one page and reflows it. If the page is incomplete then the page sequence
creates a continuation for the page and sets it as the page's next sibling (it is also the page's
@ -455,7 +661,7 @@
<a href="#s20">&lt;</a>
<a href="#s22">&gt;</a>
</div>
<h1>Intro to paganated reflow</h1>
<h1>Intro to paginated reflow</h1>
<ul>
<li>If the reflowee is incomplete, the reflower has the responsiblity to:
<ul>
@ -477,7 +683,7 @@
<a href="#s21">&lt;</a>
<a href="#s23">&gt;</a>
</div>
<h1>Intro to paganated reflow</h1>
<h1>Intro to paginated reflow</h1>
<ul>
<li>Special assumptions
<ul>
@ -547,7 +753,7 @@
<div class="slide" id="s25">
<div class="nav">
<a href="#s24">&lt;</a>
<a href="#s26">&gt;</a>
<a href="#s0">&gt;</a>
</div>
<h1>Table paginated reflow</h1>
<ul>
@ -568,215 +774,6 @@
</ul>
</div>
<div class="slide" id="s26">
<div class="nav">
<a href="#s25">&lt;</a>
<a href="#s27">&gt;</a>
</div>
<h1>Managing style contexts</h1>
<ul>
<li>Style contexts must (in most cases) be created before frames are
constructed, to determine what frame to create.</li>
<li>Parent style context determines inheritance; it should
always be the content parent. [design flaw in frame/SC
relationship]</li>
<li>Three functions for creating style contexts on
<code>nsIStyleSet</code>, wrapped by similarly named ones on
<code>nsIPresContext</code>:
<ul>
<li><code>ResolveStyleContextFor</code>: For elements.</li>
<li><code>ResolvePseudoStyleContextFor</code>: for
pseudo-elements (<code>:first-letter</code>,
<code>:before</code>, etc.)</li>
<li><code>ResolveStyleContextForNonElement</code>: skips rule
matching and uses root rule node (text frame optimization)</li>
</ul>
</li>
</ul>
</div>
<div class="slide" id="s27">
<div class="nav">
<a href="#s26">&lt;</a>
<a href="#s28">&gt;</a>
</div>
<h1>Managing style contexts</h1>
<ul>
<li>Style context resolving functions will walk the rule processors
in <code>StyleSetImpl::FileRules</code>, find the correct rule
node, and find a current child of the parent (&ldquo;sibling
sharing&rdquo;) or create a new child.</li>
<li>Style context doesn't hold pointer to content, just rule
node.</li>
</ul>
</div>
<div class="slide" id="s28">
<div class="nav">
<a href="#s27">&lt;</a>
<a href="#s29">&gt;</a>
</div>
<h1>Dynamic changes to content</h1>
<ul>
<li><code>FrameManager::ReResolveStyleContext</code> destroys and
recreates style contexts for existing frames (rule node pointer
immutable).</li>
<li><code>ReResolveStyleContext</code> is messy because it needs to
create and parent style contexts correctly (sibling
sharing may not be the same) rather than just changing data.
[design flaw, again]</li>
<li>Any specially-parented style contexts (not along <em>frame</em>
parents, which need not be content parents) are reconstructed
using <code>nsIFrame::GetParentStyleContextFrame</code>.</li>
<li>Can return same style context due to sibling sharing unless
we're destroying the rule tree for a style sheet/rule
removal.</li>
</ul>
</div>
<div class="slide" id="s29">
<div class="nav">
<a href="#s28">&lt;</a>
<a href="#s30">&gt;</a>
</div>
<h1>Dynamic changes to content</h1>
<ul>
<li><code>ReResolveStyleContext</code> calculates differences
(repaint, reflow, reframe, etc.) between style old and new style
contexts and does appropriate cleanup</li>
<li>It uses <code>nsIStyleContext::CalcStyleDifference</code>, which
only computes differences for structs that have been
requested. (I'll call this the data-struct-based hint
mechanism.)</li>
<li>Caller of <code>nsIFrameManager::ComputeStyleChangeFor</code>
processes the change list, which has been built to avoid
duplication.</li>
<li>We also have <code>ReParentStyleContext</code>, used in a few
places (usually during frame construction), but it's broken (has
many bugs that <code>ReResolveStyleContext</code> <em>used
to</em> have).</li>
</ul>
</div>
<div class="slide" id="s30">
<div class="nav">
<a href="#s29">&lt;</a>
<a href="#s31">&gt;</a>
</div>
<h1>Dynamic changes to content: optimizations</h1>
<ul>
<li>We optimize <em>attribute changes</em> by storing all the
attributes that have an effect on which rules match and only
doing a <code>ReResolveStyleContext</code> if the attribute has
an effect. <code>nsIStyleSheet::AttributeAffectsStyle</code>
(should be on <code>nsIStyleRuleProcessor</code>).</li>
<li>We optimize <em>event state changes</em> (<code>:hover</code>,
<code>:active</code>, etc.) using
<code>nsIStyleRuleProcessor::HasStateDependentStyle</code>,
which is much more accurate. The CSSRuleProcessor
implementation does a slightly modified form of selector
matching to implement it (includes matching on the middle of
selectors to catch <code>p:hover a</code>).</li>
</ul>
</div>
<div class="slide" id="s31">
<div class="nav">
<a href="#s30">&lt;</a>
<a href="#s32">&gt;</a>
</div>
<h1>Style attribute changes</h1>
<ul>
<li>We handle <code>style</code> attribute (&ldquo;inline
style&rdquo;) changes in a different
way from other changes to style rules.</li>
<li>As for other style changes, we have to &ldquo;walk&rdquo; the
rule tree and clear all the style data coming from the old
inline style <code>nsIStyleRule</code>, since there could be an
<code>!important</code> rule that overrides it, which would
allow dynamic changes to put the style attribute in multiple
places in the rule tree. However, we maintain a hashtable just
for inline style rules so that we don't have to walk the whole
tree to find the nodes.</li>
<li><code>nsCSSFrameConstructor::AttributeChanged</code> only
reresolves style on the subtree of the element, just like other
<em>attribute</em> changes.</li>
<li>Different hint mechanism (from rule structs, not data structs)
could make <code>AttributeChanged</code> just go straight to a
framechange, instead.</li>
<li>Bugs due to failure to call
<code>nsIFrame::DidSetStyleContext</code>.</li>
</ul>
</div>
<div class="slide" id="s32">
<div class="nav">
<a href="#s31">&lt;</a>
<a href="#s33">&gt;</a>
</div>
<h1>Style sheet addition and removal</h1>
<ul>
<li>Handled in pres shell.</li>
<li><code>PresShell::ReconstructStyleData</code> calls
<code>FrameManager::ComputeStyleChangeFor</code>
(ReResolve) and then processes the framechange list.</li>
<li>Rebuilds rule tree if stylesheet was removed to avoid dangling
pointers (and perhaps aliasing that would cause problems).
Otherwise we'd have to walk rule tree and compare each rule node
to every rule in the sheet (<code>O(rules *
rule-nodes)</code>).</li>
<li>When rebuilding rule tree, we have to clear cached style
contexts from XUL menus and trees.</li>
</ul>
</div>
<div class="slide" id="s33">
<div class="nav">
<a href="#s32">&lt;</a>
<a href="#s34">&gt;</a>
</div>
<h1>Style rule changes</h1>
<ul>
<li>Handling of style rule changes is done in frame constructor
(called from style set, called from pres shell, which is a
document observer) and in the pres shell. Code should be
merged.</li>
<li>Rule change applies the rule-struct hint as if the rule matched
the root element. (inefficient) It does clearing of style data
(through <code>StyleSetImpl::ClearStyleData</code>) by walking
the rule tree and then the style context tree.
(could be handled by simultaneous clearing and
difference calculation of data (somewhat tricky))</li>
<li>Rule addition and removal just rebuild the entire world. We
could at least do what we do for sheet addition/removal, or
slightly better, by searching the rule tree (only one rule this
time) instead of rebuilding it.</li>
<li><strong>Lots of room for optimization here.</strong> (but beware
<code>DidSetStyleContext</code>)</li>
</ul>
</div>
<div class="slide" id="s34">
<div class="nav">
<a href="#s33">&lt;</a>
<span style="background: white; color: white">&lt;</span>
</div>
<h1>The style system</h1>
<center>
Style sheets &amp; rules<br>&darr;<br>
Rule tree<br>&darr;<br>
Style context interface
</center>
</div>
</body>
</html>