[XSL-LIST Mailing List Archive Home] [By Thread] [By Date]

Re: [xsl] Getting years from duration

Subject: Re: [xsl] Getting years from duration
From: Martin Holmes <mholmes@xxxxxxx>
Date: Sat, 8 Jun 2013 16:55:52 -0700

We only wanted years, but a plain subtraction won't do, as others have noted; you need to check whether to subtract a year from the result or not, depending on the times of year of the births and deaths. So:

Born 1950 December 31
Died 2000 January 1

The plain subtraction will give you 50, but the age is actually 49 (plus a day). [If my math is correct at the end of a LONG day.]


On 13-06-08 01:42 PM, Michele R Combs wrote:
Do you need the age down to month and day? If not, rather than manipulating the attribute value as a date per se, can you just take the first four characters (the year) of each and do a simple subtraction?

From: Martin Holmes [mholmes@xxxxxxx]
Sent: Friday, June 07, 2013 11:26 PM
To: xsl-list@xxxxxxxxxxxxxxxxxxxxxx
Subject: [xsl] Getting years from duration

We came up against what looked like a very simple XPath issue today, and
hit a brick wall with it. Given data that looks like this:

          <person role="author">
            <persName>Milton, John</persName>
            <birth when="1600-12-09"/>
            <death when="1674-11-08"/>

we want to calculate the age of the person at death. So we thought:
subtract the death date from the birth date to get a duration, then
extract the years from the duration:

<xsl:template match="person">

      <xsl:variable name="life" select="xs:date(death/@when) -

<xsl:variable name="age" select="years-from-duration($life)"/>

      <xsl:text>Age at death: </xsl:text>
      <xsl:value-of select="$age"/>


However, the only value we were able to get back, after trying all manner of permutations and casts, was zero. It appears that what comes back from the date subtraction (which I think uses the op:subtract-dates() operator) is always an xs:dayTimeDuration, and that cannot AFAIKS be manipulated into anything from which a year can be extracted.

It's always possible to get the number of days and divide by 365.25, but
it seems strange to have to do that. Given the range of date- and
duration-related functions, I'm sure there must be some better way of
getting the result. Does anyone know?

This is using Saxon



-- Martin Holmes University of Victoria Humanities Computing and Media Centre (mholmes@xxxxxxx)

Current Thread