[xquery-talk] a conceptual clarification

[Michael Rys]

It is not quite the same. Variables in declarative/functional languages
are different from imperative/procedural languages; especially inside an
iterator. 

 

Whether variables should be redeclarable is kind of a religious issues
among the functional language crowd (well, variables are). I think that
re-assignment is ok if people understand the functional processing
paradigm. But since many currently do not understand it and since the
W3C has not closed the spec yet, I am leaning towards erroring now and
relaxing the error later...

 

Best regards

Michael

 

> -----Original Message-----

> From: talk-bounces at xquery.com [mailto:talk-bounces at xquery.com] On
Behalf

> Of Bas de Bakker

> Sent: Friday, March 04, 2005 4:41 AM

> To: Amitabh Ojha

> Cc: talk at xquery.com

> Subject: Re: Re : [xquery-talk] a conceptual clarification

> 

> Shadowing a variable with another variable with the same name is

> possible in many languages, including Java.  It is also generally

> confusing and therefore considered bad practice (although I do not
agree

> that it should be disallowed).  In this respect there is nothing
special

> about XQuery.

> 

> I fail to see why you would consider using a let statement within a
for

> statement to be confusing, nor do I see a better alternative.  This is

> just like

> 

> for (int i = 0; i < n; ++i) {

>      Object object = a[i];

>      // Use object

> }

> 

> in Java, which is an extremely common way of programming in many

> languages.

> 

> Regards,

> Bas de Bakker

> 

> 

> Amitabh Ojha wrote:

> > Dear Sir,

> >

> > It has been a very useful exchange of  messages/ replies from
experts on

> > this theme. However, no discussion has taken place as yet on  one
point

> > which I had flagged earlier.  I elaborate it here.

> >

> > It is evident now  that  it is a bad practice to have two let

> > statements, one after the other, define a variable  with the same
name,

> > as  is  cited  in the following example  (even if it is not illegal
as

> > per XQuery Specs)

> >

> > let $x := ......

> > let $x := ......

> >

> > Following from this, will it be correct to assume that  the
appearance

> > of a let statement after a for statement be also treated as equally

> > undesirable and error-prone programming technique in XQuery (because
in

> > course of  iteration,  a let statement is then forced to rebind  the

> > same variable name say  $x several times, probably forcing it to
bind to

> > a new value each time).  I cite two examples here. In case of
my:foo ()

> > the result is same as one would expect in Java i.e  1, 2, 3, 4, 5.
But

> > in case of  my:bar () we get 101, 102, 103, 104, 105 - this is not
what

> > a Java programmer will expect (although in XQuery context one knows
why

> > it is so).

> >

> > declare function my:foo ()

> > {

> > for $i in (1 to 5)

> > let $x := $i

> > return $x

> > };

> >

> > declare function my:bar ()

> > {

> > let $x := 100

> > for $i in (1 to 5)

> > let $x :=  $x  +  $i

> > return $x

> > };

> >

> > Regards.

> >

> > Amitabh Ojha

> _______________________________________________

> talk at xquery.com

> http://xquery.com/mailman/listinfo/talk

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://xquery.com/pipermail/talk/attachments/20050304/32f34777/attachment-0001.htm