I used system-property() to query for vendor, version and vendor-URL
and this is what I get:
Vendor is Apache Software Foundation-
1-
http://xml.apache.org/xalan-j
So I'm using xalan, XSLT 1.0, after all.
Michael, if you don't mind me asking, how do you explicitly
set the precision? It's not something to do with xs:float() and
xs:double() is it? I tried them out but it seems they're not
supported.
Oh and as I mentioned, I only got this on SunOS. Would it have
something to do with the OS? How can I tell if it's something
inherent in that machine (how it processes floating-point in
general) and not XSL processor or Java?
Thanks,
On 26 Jan 2006 at 8:49, Michael Kay wrote:
I agree, this result is way out. It looks to me suspiciously as if
it's carried out using single-precision rather than double-precision
floating point.
Saxon (both 6.5.5 and 8.6.1) give
<?xml version="1.0" encoding="utf-8"?>6.4_
3000_
19200_
dom4j doesn't include an XSLT processor of its own, as far as I know.
Use xsl:system-property('xsl:vendor') to find out which processor you
are using.
I changed the code to explicitly use single-precision float and the
result was
<?xml version="1.0" encoding="UTF-8"?>6.4_
3000_
19200.000286102295_
which is still closer than the result you are seeing.
Michael Kay
http://www.saxonica.com/
-----Original Message-----
From: Thang Le [mailto:tle(_at_)neptium(_dot_)com]
Sent: 26 January 2006 01:16
To: xsl-list(_at_)lists(_dot_)mulberrytech(_dot_)com
Subject: [xsl] Floating point precision error
This is something I don't understand. Given this XSL snippet:
<xsl:value-of select="$nodes[1]/unitamount" />_
<xsl:value-of select="$nodes[1]/quantity" />_
<xsl:value-of select="($nodes[1]/quantity * $nodes[1]/unitamount)"
/>_
I'm getting the printout as:
6.4_
3000_
19199.908447265625_
The literal values in XML is 6.4 and 3000. If I round the result to
2 decimal places, I still get 19199.91, a far cry from what it
should be (19200).
I'm using dom4j 1.6, so I guess that's XSLT 1.0? Also this
only happens
on SunOS and not on Windows. Still, the error is huge and obvious.
Isn't there a way to force the calculation to be more exact?
Thanks,
Thang Le
--
NeoMail - http://neomail.sourceforge.net
--~-----------------------------------------------------------------
- XSL-List info and archive:
http://www.mulberrytech.com/xsl/xsl-list To unsubscribe, go to:
http://lists.mulberrytech.com/xsl-list/ or e-mail:
<mailto:xsl-list-unsubscribe(_at_)lists(_dot_)mulberrytech(_dot_)com> --~--
--~------------------------------------------------------------------
XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list
To unsubscribe, go to: http://lists.mulberrytech.com/xsl-list/ or
e-mail: <mailto:xsl-list-unsubscribe(_at_)lists(_dot_)mulberrytech(_dot_)com>
--~--
--~------------------------------------------------------------------
XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list
To unsubscribe, go to: http://lists.mulberrytech.com/xsl-list/
or e-mail: <mailto:xsl-list-unsubscribe(_at_)lists(_dot_)mulberrytech(_dot_)com>
--~--