Menu
▾
▴
Re: [Gnuplot-bugs] 2 bugs (more detailed description of problem)
|
From: Florian H. <e00...@st...> - 2007-05-04 12:52:54
|
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
<title></title>
</head>
<body bgcolor="#ffffff" text="#000000">
Message: 7<br>
<blockquote
cite="mid...@li..."
type="cite">
<pre wrap="">Date: Thu, 3 May 2007 14:38:44 -0700
From: Ethan Merritt <a class="moz-txt-link-rfc2396E" href="mailto:merritt@u.washington.edu"><merritt@u.washington.edu></a>
Subject: Re: [Gnuplot-bugs] 2 bugs (more detailed description of
problem)
To: <a class="moz-txt-link-abbreviated" href="mailto:gnu...@li...">gnu...@li...</a>
Cc: Hans-Bernhard Br?ker <a class="moz-txt-link-rfc2396E" href="mailto:HBB...@t-..."><HBB...@t-...></a>, Bug reports for
gnuplot <a class="moz-txt-link-rfc2396E" href="mailto:gnu...@li..."><gnu...@li...></a>
Message-ID: <a class="moz-txt-link-rfc2396E" href="mailto:200705031438.44696.merritt@u.washington.edu"><200705031438.44696.merritt@u.washington.edu></a>
Content-Type: text/plain; charset="iso-8859-1"
On Thursday 03 May 2007 13:32, Hans-Bernhard Br?ker wrote:
</pre>
<blockquote type="cite">
<blockquote type="cite">
<pre wrap="">I bet you are using a platform independent library for the manipulation
of floats ...
</pre>
</blockquote>
<pre wrap="">Based on what clue?
</pre>
</blockquote>
</blockquote>
Clue: It is definitely a floating point range (at least the initial
plotting region is specified<br>
using floating point constants only). Why does it complain about<br>
integers than??? Why does it depend on the architecture dependent<br>
*size* of *integers* if I am dealing with floating point numbers only.<br>
So floats are never converted (internally) to integers in gnuplot?<br>
<blockquote
cite="mid...@li..."
type="cite">
<blockquote type="cite">
<pre wrap="">
</pre>
<blockquote type="cite">
<pre wrap="">(where does the non gnuplot-style error message "integer overflow;
change to floating point"
come from, why does the error message differ from platform to platform?).
</pre>
</blockquote>
<pre wrap="">Because we're *not* using any such library. If we did, the message
would be the same, and the behaviour would be platform-independent.
</pre>
</blockquote>
<pre wrap=""><!---->
</pre>
</blockquote>
<blockquote
cite="mid...@li..."
type="cite">
<pre wrap="">The error message comes from scanner.c line 285, and it means exactly
what it says. Different platforms (or different compiler options) may
yield different integer sizes. Obviously a 16-bit integer will overflow
long before a 64-bit integer.
</pre>
</blockquote>
The fascinating thing is that the error messages are in fact
*different*.<br>
Only on a 64bit system (debian, gcc compiled) it complains about
integers<br>
for obscure reasons. On other system gnuplot worries about y-ranges.<br>
<blockquote
cite="mid...@li..."
type="cite">
<pre wrap="">I can't seem to read the original reports via SourceForge, due to
"500 - Internal Server Error" from the SourceForge server. Are you
getting this error message as a result of interactive mouse zooming?
</pre>
</blockquote>
Yes, correct.<br>
</body>
</html>
|
