[css3-layout] Mentioned the suggested 'dense' keyword for the

automatic placement algorithm as an issue.

--HG--
extra : rebase_source : f6e2b6a6dd58a7a3707bd869c9c1beb4ae0f4f9a

Author: bert-github

css-template/Overview.src.html

@@ -2224,7 +2224,8 @@ <h3>Stacking elements on top of a grid: 'grid-area' and 'grid-auto'</h3>
 <table class=propdef>
 <tbody>
 <tr><th>Name: <td><dfn>grid-auto</dfn>
-<tr><th><a href="#values" >Value</a>: <td>none | rows | columns
+<tr><th><a href="#values" >Value</a>: <td>none
+| [ rows && dense? ] | [ columns && dense? ]
 <tr><th>Initial: <td>none
 <tr><th>Applies to: <td>elements that have a <span>grid
 ancestor</span> and whose 'position' is equal to 'grid'
@@ -2350,16 +2351,28 @@ <h3>Stacking elements on top of a grid: 'grid-area' and 'grid-auto'</h3>
 
 <dt>rows
 <dd>
-<p>The element is positioned as high as possible in the grid (and if
-there are multiple solutions at the same height: as far to the left as
-possible) such that is doesn't overlap with any earlier elements in
-the document that have the same <em>grid ancestor</em> and 'grid-area'
-other then ''none''. More precisely: the element is placed as for
-'grid-area: <var>x</var> <var>y</var> <var>W</var> <var>H</var>', where
-<var>x</var> and <var>y</var> are chosen as follows: <span
-class=issue>Define. Tight packing (i.e., closest to top left corner)
-or preserve element order (i.e., to the right or below of previous
-element)?</span>
+<p>Loosely: the element is positioned in the currently last row, to
+the right of all elements in the row, unless it doesn't fit, in which
+case it is positioned in the first column of a newly created row.
+
+<p>More precisely: let (<var>x</var><sub><var>i</var></sub>,
+<var>y</var><sub><var>i</var></sub>,
+<var>W</var><sub><var>i</var></sub>,
+<var>H</var><sub><var>i</var></sub>) be the columns, rows, widths and
+heights of element <var>i</var> for all elements that come earlier in
+the document and that are positioned in the current grid with
+'grid-area'. Let <var>L</var> =
+max<sub><var>i</var></sub>(<var>y</var><sub><var>i</var></sub>). If
+there are no <var>i</var>, then <var>L</var> = 1. One can think of
+<var>L</var> as the row that is currently being filled. Let
+<var>C</var> = max<sub>j</sub>(<var>x</var><sub><var>j</var></sub> +
+<var>W</var><sub><var>j</var></sub>) for all <var>j</var> such that
+<var>y</var><sub><var>j</var></sub> = L. If there are no <var>j</var>,
+then <var>C</var> = 1. This is the next available column in the
+current row. Then the element is positioned at (<var>L</var>,
+<var>C</var>) if <var>C</var> = 1 or <var>C</var> + <var>W</var> is
+less then the <em>number of columns</em> and it is positioned at
+(<var>L</var> + 1, 1) otherwise.
 
 <p>The name 'rows' comes from the fact that this algorithms tends to
 fill a row with subsequent elements and then adds new rows
@@ -2369,14 +2382,12 @@ <h3>Stacking elements on top of a grid: 'grid-area' and 'grid-auto'</h3>
 
 <dt>columns
 <dd>
-<p>The element is positioned as far to the left as possible in the
-grid (and if there are multiple solutions: as high as possible) such
-that is doesn't overlap with any earlier elements in the document that
-have the same <em>grid ancestor</em> and 'grid-area' other then
-''none''. More precisely: the element is placed as for 'grid-area:
-<var>x</var> <var>y</var> <var>W</var> <var>H</var>', where
-<var>x</var> and <var>y</var> are chosen as follows: <span
-class=issue>define</span>
+<p class=issue>Analogous…
+
+<dt>dense
+<dd>
+<p>The element is positioned in the first available unused space in
+the grid. <span class=issue>Drop? or define?</span>
 </dl>
 
 <p class=issue>The pair of keywords ''same/next'' and the pair