<html>
<head>
<meta content="text/html; charset=utf-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix">Le 17/01/2018 à 18:16, antoine
monmayrant a écrit :<br>
</div>
<blockquote cite="mid:1834a2f4-8cad-da2a-55c8-4a2496a56869@laas.fr"
type="cite">Hello all,
<br>
<br>
<br>
I might say something really stupid, but can't we define a special
variable like %eps, %i, etc to act as a black hole ?
<br>
Something like %null (like piping to /dev/null or a null pointer)
or a better name (%blackhole, %lostforever, %byebye, ;-) ).
<br>
The idea being that no memory is allocated when someone try to
affect some value to it:
<br>
<br>
[%null,%null,kb] = intersect(grand(1,10,"uin",0,9),
grand(1,10,"uin",0,9));
<br>
</blockquote>
<br>
A true black hole has even no name :)<br>
Otherwise, the shorter the better, and "_" would be nice after
deprecating _() (*).<br>
<br>
Anyway, IMO this black hole feature has a very low priority compared
to some other ones, like some big regressions (bug <a
href="http://bugzilla.scilab.org/show_bug.cgi?id=13808">13808</a>
= <a href="http://bugzilla.scilab.org/show_bug.cgi?id=14881">14881</a>,
sparse() with repeated indices, etc etc) or some very and longly
expected ones like a varprot()...<br>
<br>
Samuel<br>
<br>
(*) it could be renamed something like _tr() (standing for <i>tr</i>anslate).<br>
The number of occurrences of _() to be converted in the Scilab
native code is very important, but the conversion might be automated
(since the gettext() scanner has a reliable detection criterium, and
only .sci, .sce, .tst and .dia.ref files are targeted).<br>
gettext() and _() are almost unused in external modules, because the
gettext domain feature has never worked (or please just give us a
working example... Never seen a single one).<br>
<br>
</body>
</html>