<html>
<head>
<meta content="text/html; charset=utf-8" http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<div class="moz-cite-prefix">Hello Clément,<br>
<br>
Le 09/06/2016 19:08, Clément David a écrit :<br>
</div>
<blockquote
cite="mid:1465492106.10936.9.camel@scilab-enterprises.com"
type="cite">
<pre wrap="">Dear Scilab contributors,
For the (near to be released) beta-2, I updated the CHANGES files to use the Markdown format [1].
The idea is to be able to generate some document (html and pdf) from it in a semi-automatic way.
Another point would be to have it nicely formatted on hosting platforms (github/gitlab) that render
markdown on the fly.
What do you think about that ? It is currently not clear to me if we should add an empty line
between each `*` and if we should use set bugzilla URLs.
[1]: <a class="moz-txt-link-freetext" href="https://codereview.scilab.org/#/c/18222">https://codereview.scilab.org/#/c/18222</a>
Thanks for your remark,
PS: for the beta-2, we will probably push it but any modification is possible for the stable
release.</pre>
</blockquote>
<br>
My feeling and suggestions are the following:<br>
<ul>
<li>a formated output is certainly better than a simple-text one
:)</li>
<li>I found the former Scilab releases notes very smart. The
presentation on 2 columns was smart and welcome. It allows using
a smaller font, to get more compact but still very readable
notes. <br>
</li>
<li>A sans-serif font is always more readable. Arial is nice.</li>
<li>An hypertext output is nice also (or mainly) because it allows
hyperlinks, including ones targeting bugzilla. However, any
text-decoration should be avoided: underlining decrease links
readability.<br>
<br>
</li>
<li>To avoid heaviness of the bug lists, heading dots may be
removed. <br>
<br>
</li>
<li>In CHANGES, <br>
</li>
<ul>
<li>In the final HTML rendering of the bugs lists, i find that
Pierre-Aimé proposal to remove "fixed" is indeed better.<br>
</li>
<li>short references to bugzilla should be preferred to leave
more space to the bug description: <br>
<a class="moz-txt-link-freetext" href="http://bugzilla.scilab.org/2104">http://bugzilla.scilab.org/2104</a> instead of
<a class="moz-txt-link-freetext" href="http://bugzilla.scilab.org/show_bug.cgi?id=2104">http://bugzilla.scilab.org/show_bug.cgi?id=2104</a></li>
<li>Without what you call "markdown", i definitely preferred
noting functions with parentheses like in "linespace()"
instead of only "linespace". The purpose was to highlight
Scilab reserved words wrt to common words in a sentence.
However, if quoting markdown are used, i think that
parentheses are no longer required: They would now make
notations heavier than required. So i propose to not use them
in forthcoming CHANGES.<br>
<br>
</li>
</ul>
<li>For compactness, i definitely prefer avoiding any \n after
each `*` item</li>
<li>For the same reason, i think that any sublist having only one
term per line shall be avoided. An inline list shall be
preferred. For instance,</li>
<ul>
<li>Symbolic module functions have been removed:
<ul>
<li><code>addf()</code></li>
<li><code>cmb_lin()</code></li>
<li><code>ldivf()</code></li>
<li><code>mulf()</code></li>
<li><code>rdivf()</code></li>
<li><code>solve()</code></li>
<li><code>subf()</code></li>
<li><code>trianfml()</code></li>
<li><code>trisolve()</code></li>
<li><code>block2exp()</code></li>
</ul>
<code>looks useless. I would prefer much more simply</code></li>
<li>
<p>Symbolic module functions have been removed:<code> </code><tt>addf</tt><code>,
</code><tt>cmb_lin</tt><code>, </code><tt>ldivf</tt><code>,
</code><tt>mulf</tt><code>, </code><tt>rdivf</tt><code>, </code><tt>solve</tt><code>,
</code><tt>subf</tt><code>, </code><tt>trianfml</tt><code>,
</code><tt>trisolve</tt><code>, block2exp</code></p>
</li>
</ul>
<li>As well, </li>
<ul>
<li>
<p>Functions based on former Scilab stack have been removed:</p>
<ul>
<li><code>comp()</code></li>
<li><code>errcatch()</code></li>
<li><code>iserror()</code></li>
<li><code>fun2string()</code></li>
<li><code>getvariablesonstack()</code></li>
<li><code>gstacksize()</code></li>
<li><code>macr2lst()</code></li>
<li><code>stacksize()</code></li>
<li><code>code2str()</code></li>
<li><code>str2code()</code></li>
<li><code>-mem</code> launching option (used to set <code>stacksize()</code>
at startup).</li>
</ul>
<br>
could be abstracted in</li>
<li>
<p>Functions based on former Scilab stack have been removed:<code></code><code></code></p>
</li>
<ul>
<li>
<p><code></code><code>comp, </code><code>errcatch,</code><code>
iserror</code><code>, fun2string,</code><code>
getvariablesonstack, </code><code>gstacksize, </code><code>macr2lst,
</code><code>stacksize, </code><code>code2str, </code><code>str2code</code><code></code></p>
</li>
<li>
<p><code>-mem</code> launching option (used to set <code>stacksize()</code>
at startup).</p>
</li>
</ul>
</ul>
</ul>
<p>I have implemented some of these suggestions in the attached
CHANGES.html:</p>
<ul>
<li>in <head><style>:<br>
a:link { text-decoration: none; font-weight: bold;}<br>
body { font-family:"arial"; font-size:11px }<br>
ul { page-break-before:avoid; } /* not enough to
actually avoid page breaks at bad places. A pity. */<br>
h2 { page-break-after:avoid; } /* .... */<br>
h3 { page-break-after:avoid; } /* .... */<br>
<br>
</li>
<li>The list of bugs on 6.0.06beta2 release is encapsulated in <br>
<div style="column-count:2; -moz-column-count:2;
-webkit-column-count:32 "><br>
<ul style="list-style-type: none;"><br>
...<br>
</ul><br>
</div><br>
</li>
<li>The second list is left as is for comparison.<br>
<br>
</li>
</ul>
I don't know if such options could be implemented if you think they
could improve the output.<br>
<br>
I was expecting other contributions before / instead spending my
night to write to you here.<br>
<br>
But apparently, considerations about subtracted uniform
distributions are much more efficient to awake everybody on Scilab
mailing lists.<br>
Next time, you shall introduce your call-for-comments with something
about maths :))<br>
<br>
Good night<br>
Samuel<br>
<br>
</body>
</html>