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

Re: [xsl] Things that make you go Hmmmm!

Subject: Re: [xsl] Things that make you go Hmmmm!
From: "Abel Braaksma (Exselt)" <abel@xxxxxxxxxx>
Date: Sat, 29 Mar 2014 16:40:04 +0100

On 29-3-2014 16:11, Michael Sokolov wrote:
> On 3/29/2014 10:43 AM, Abel Braaksma (Exselt) wrote:
>> While a tempting idea, it wouldn't work in practice I think, even if,
>> for a moment, we ignore backward compatibility issues. Consider the
>> identity template, the deepest copy always ends up with an empty
>> selection in the xsl:apply-templates. This is effectively the same as an
>> empty sequence constructor. Your proposal would mean that:
>> <xsl:copy />
>> and
>> <xsl:copy>
>>     <xsl:apply-templates select="()" />
>> </xsl:copy>
>> would both behave differently, depending on the context node. The first
>> would copy the whole subtree without further processing, the second,
>> which also has an effectively empty sequence constructor, would only
>> copy the current node without its children.
> I don't think I understand this.  Are you saying that <xsl:template
> match="@*|node()"> matches the empty sequence? That would be quite
> surprising to me.  I can see that <xsl:apply-templates /> on an empty
> node could cause an attempt to match an empty sequence, but does that
> actually fire rules?

No, not directly. I meant to say that in the modified identity template
example that I gave, the processor will recursively go through all
nodes, and eventually will reach an attribute (in which case "@*|node()"
will select nothing) or a leaf node like text() or
processing-instruction() (in which case it will also select nothing).
The result of those deepest iterations is that there is nothing to copy
anymore and xsl:copy will do nothing.

My point was that it might lead to surprises if such a statement that
does not select anything, is used directly. Hence my select="()". In the
event where we were to allow your proposal of having xsl:copy also
behave as xsl:coy-of, I think this case will lead to problems in
practice, because you effectively say "when xsl:copy has no sequence
constructor, copy everything, when it has an empty sequence constructor
or one that selects an empty node set, copy nothing".

> Yeah, I think that priniciple is over-stressed - as I said in an
> earlier post, I think judicious suprises can be an essential spur to
> learning, but I recognize this is a matter of taste.

True, and no language in existence will ever give the same surprises
everywhere and to everyone. People come to a language with different
backgrounds and different use-cases, leading to different learning
curves. and hence different surprises. It is up to the language
designers to try to make them as few as possible, but there always comes
a time when someone thinks there are too many of those surprises, or in
the wrong spots of the language. Which in turn, keeps us sharp ;).


Abel Braaksma
Exselt XSLT 3.0 processor

Current Thread