RE: [xsl] Reference to variable cannot be resolved.

Subject: RE: [xsl] Reference to variable cannot be resolved.
From: "Martinez, Brian" <brian.martinez@xxxxxxxx>
Date: Fri, 14 Feb 2003 10:18:21 -0700
> From: Jeni Tennison [mailto:jeni@xxxxxxxxxxxxxxxx]
> Sent: Friday, February 14, 2003 9:43 AM
> Subject: Re: [xsl] Reference to variable cannot be resolved.
> Hi Brian,
> > The example given in the spec seems a bit misleading, because it
> > really illustrates the restrictions of scope and not the practice of
> > shadowing. The following is illegal in either spec:
> >
> > <xsl:template match="/">
> >   <xsl:variable name="x" select="1"/>
> >   <xsl:variable name="x" select="$x+1"/>
> >   <xsl:value-of select="$x"/>
> > </xsl:template>
> Um, I think that's legal in XSLT 2.0. What makes you think that it
> isn't?

Uh . . . because I didn't read the following paragraph in Section 9.7 very
closely?  ;-)

"It is also not an error if a binding established by a local xsl:variable or
xsl:param element shadows another binding established by another local
xsl:variable or xsl:param. However, such shadowing is discouraged and
implementations may output a warning when it occurs."

However, this illustrates why I think the code example given is misleading,
because it doesn't really illustrate the concept of shadowing a local
variable, but rather the restrictions placed on its scope: the variable
declaration inside the for-each will never override the outer variable with
the same name.

Still, I'm very surprised that my above example would be legal, and it makes
Dimitre's objection that much clearer: if the value of $x can't be changed
by a subsequent binding within the same scope, then why allow the practice
at all?

> I think that the argument for allowing shadowing of local variables
> was that it allows those unimaginative programmers who can only think
> of the one variable name (you know who I'm talking about, David C.) to
> reuse that variable name while carrying out several steps of
> processing on the same value, precisely as you have above.
> Note that both XSLT 1.0 and XSLT 2.0 allow you to do:
> <xsl:template match="/">
>   <xsl:variable name="x">
>     <xsl:variable name="x" select="1" />
>     <xsl:value-of select="$x + 1" />
>   </xsl:variable>
>   <xsl:value-of select="$x" />
> </xsl:template>

True . . . but it does makes sense, since the inner $x is not in the same
scope as the outer $x.  What I'm struggling to understand is why the spec
would allow a programmer even to attempt to assign a new value to $x.


| brian martinez                              brian.martinez@xxxxxxxx |
| senior gui programmer                                  303.708.7248 |
| trip network, inc.                                 fax 303.790.9350 |
| 6436 s. racine cir.                             englewood, co 80111 |
|          |

 XSL-List info and archive:

Current Thread