Hi, That would base the type annotation on the lexical value. For instance, the string '0' would be labelled as "integer" (given you test for xs:integer before xs:string): let $var := '0' return if ( $var castable as xs:integer ) then 'integer' else if ( $var castable as xs:string ) then 'string' else 'unknown type' (: -> 'integer' :) The operator "instance of" would solve that problem. But then, for that purpose, there is the "typeswitch" instruction: typeswitch ( '0' ) case xs:integer return 'integer' case xs:string return 'string' default return 'unknown type' (: -> 'string' :) Regards, -- Florent Georges http://fgeorges.org/ http://h2oconsulting.be/ On 29 July 2015 at 17:34, Eliot Kimber <ekimber@contrext.com> wrote:

I believe you need to use the "castable as" or "instance of" operators to determine the type, e.g.:

let $typeName := if ($somevar castable as xs:integer) then 'integer' else if ($somevar castable as xs:string) then 'string' else 'unknown type'

Or something close to that. There is also "instance of", which might be

I also found this paper, which might be interesting:

http://www.balisage.net/Proceedings/vol8/html/Holstege01/BalisageVol8-Holst ege01.html

Cheers,

E. ---- Eliot Kimber, Owner Contrext, LLC http://contrext.com

On 7/29/15, 9:54 AM, "Rob Stapper" <basex-talk-bounces@mailman.uni-konstanz.de on behalf of r.stapper@lijbrandt.nl> wrote:

...

Hello,

I must be overlooking something, so excuses in advance, but: How can I retrieve the datatype of a variable?

TIA, Rob Stapper

________________________________________

<https://www.avast.com/antivirus>

Dit e-mailbericht is gecontroleerd op virussen met Avast antivirussoftware.

www.avast.com <https://www.avast.com/antivirus>

Eliot, Florent, Thank you both for your reaction. I've been thinking about the typeswitch-command too but it feels weird: I have this variable, inside XQuery its type is known but for retrieving it I have to go through all types and ask XQuery if it is this one. I hoped for a more direct solution. I've been thinking about the inspect-function too but that one is more mend for documentation-purposes and I guess performance is an issue there. Thanx again, Rob -----Oorspronkelijk bericht----- Van: Eliot Kimber [mailto:ekimber@contrext.com] Verzonden: woensdag 29 juli 2015 19:56 Aan: Florent Georges CC: Rob Stapper; BaseX Onderwerp: Re: [basex-talk] question I am happy to be corrected. Cheers, E. ---- Eliot Kimber, Owner Contrext, LLC http://contrext.com On 7/29/15, 10:42 AM, "Florent Georges" <fgeorges@gmail.com on behalf of lists@fgeorges.org> wrote:

Hi,

That would base the type annotation on the lexical value. For instance, the string '0' would be labelled as "integer" (given you test for xs:integer before xs:string):

let $var := '0' return if ( $var castable as xs:integer ) then 'integer' else if ( $var castable as xs:string ) then 'string' else 'unknown type' (: -> 'integer' :)

The operator "instance of" would solve that problem. But then, for that purpose, there is the "typeswitch" instruction:

typeswitch ( '0' ) case xs:integer return 'integer' case xs:string return 'string' default return 'unknown type' (: -> 'string' :)

Regards,

-- Florent Georges http://fgeorges.org/ http://h2oconsulting.be/

On 29 July 2015 at 17:34, Eliot Kimber <ekimber@contrext.com> wrote:

...

I believe you need to use the "castable as" or "instance of" operators to determine the type, e.g.:

let $typeName := if ($somevar castable as xs:integer) then 'integer' else if ($somevar castable as xs:string) then 'string' else 'unknown type'

Or something close to that. There is also "instance of", which might be

I also found this paper, which might be interesting:

http://www.balisage.net/Proceedings/vol8/html/Holstege01/BalisageVol8- Hol st ege01.html

Cheers,

E. ---- Eliot Kimber, Owner Contrext, LLC http://contrext.com

On 7/29/15, 9:54 AM, "Rob Stapper" <basex-talk-bounces@mailman.uni-konstanz.de on behalf of r.stapper@lijbrandt.nl> wrote:

...

Hello,

I must be overlooking something, so excuses in advance, but: How can I retrieve the datatype of a variable?

TIA, Rob Stapper

________________________________________

<https://www.avast.com/antivirus>

Dit e-mailbericht is gecontroleerd op virussen met Avast antivirussoftware.

www.avast.com <https://www.avast.com/antivirus>

--- Dit e-mailbericht is gecontroleerd op virussen met Avast antivirussoftware. https://www.avast.com/antivirus

Hi Christian, I would say that datatype of variable "label" is JLabel. If I was programming in Java I guess I want Java to return the string: "JLabel" probably combined with the URI of the source were JLabel is defined. Cheers, Rob. PS. I'm no Java-programmer ( yet :-) -----Oorspronkelijk bericht----- Van: Christian Grün [mailto:christian.gruen@gmail.com] Verzonden: donderdag 30 juli 2015 9:55 Aan: Rob Stapper CC: Eliot Kimber; Florent Georges; BaseX Onderwerp: Re: [basex-talk] question Hi Rob, In that regard, XQuery is pretty similar to many other programming languages. Think e.g. of the following Java code: public JLabel label = new JLabel(); What do you think is the "data type" of the variable named "label", and how would you like to deal with it in Java? Cheers, Christian On Thu, Jul 30, 2015 at 9:41 AM, Rob Stapper <r.stapper@lijbrandt.nl> wrote:

Eliot, Florent,

Thank you both for your reaction.

I've been thinking about the typeswitch-command too but it feels weird: I have this variable, inside XQuery its type is known but for retrieving it I have to go through all types and ask XQuery if it is this one. I hoped for a more direct solution.

I've been thinking about the inspect-function too but that one is more mend for documentation-purposes and I guess performance is an issue there.

Thanx again,

Rob

-----Oorspronkelijk bericht----- Van: Eliot Kimber [mailto:ekimber@contrext.com] Verzonden: woensdag 29 juli 2015 19:56 Aan: Florent Georges CC: Rob Stapper; BaseX Onderwerp: Re: [basex-talk] question

I am happy to be corrected.

Cheers,

E. ---- Eliot Kimber, Owner Contrext, LLC http://contrext.com

On 7/29/15, 10:42 AM, "Florent Georges" <fgeorges@gmail.com on behalf of lists@fgeorges.org> wrote:

...

Hi,

That would base the type annotation on the lexical value. For instance, the string '0' would be labelled as "integer" (given you test for xs:integer before xs:string):

let $var := '0' return if ( $var castable as xs:integer ) then 'integer' else if ( $var castable as xs:string ) then 'string' else 'unknown type' (: -> 'integer' :)

The operator "instance of" would solve that problem. But then, for that purpose, there is the "typeswitch" instruction:

typeswitch ( '0' ) case xs:integer return 'integer' case xs:string return 'string' default return 'unknown type' (: -> 'string' :)

Regards,

-- Florent Georges http://fgeorges.org/ http://h2oconsulting.be/

On 29 July 2015 at 17:34, Eliot Kimber <ekimber@contrext.com> wrote:

...

I believe you need to use the "castable as" or "instance of" operators to determine the type, e.g.:

let $typeName := if ($somevar castable as xs:integer) then 'integer' else if ($somevar castable as xs:string) then 'string' else 'unknown type'

Or something close to that. There is also "instance of", which might be

I also found this paper, which might be interesting:

http://www.balisage.net/Proceedings/vol8/html/Holstege01/BalisageVol8 - Hol st ege01.html

Cheers,

E. ---- Eliot Kimber, Owner Contrext, LLC http://contrext.com

On 7/29/15, 9:54 AM, "Rob Stapper" <basex-talk-bounces@mailman.uni-konstanz.de on behalf of r.stapper@lijbrandt.nl> wrote:

...

Hello,

I must be overlooking something, so excuses in advance, but: How can I retrieve the datatype of a variable?

TIA, Rob Stapper

________________________________________

<https://www.avast.com/antivirus>

Dit e-mailbericht is gecontroleerd op virussen met Avast antivirussoftware.

www.avast.com <https://www.avast.com/antivirus>

--- Dit e-mailbericht is gecontroleerd op virussen met Avast antivirussoftware. https://www.avast.com/antivirus

--- Dit e-mailbericht is gecontroleerd op virussen met Avast antivirussoftware. https://www.avast.com/antivirus

:-) Well, I'm using, what you could call negative logic on domain-subsets. For example: - say we have a domain: "pears" containing pears and a domain: "apples" containing apples. - say we define a subset on the "apples"-domain as: "all apples from domain: "apples", except apple: <some_apple_id>, and apple: <some_other_apple_id>" - say we have a pear form the "pears"-domain with id: <some_pear_id>. - if I want to test if the apple-subset contains this pear I must check the domain, i.o.w. type, of the pear against the domain, i.o.w. type, of the apple-subset. If I don't the answer to this question would be: "true", which is wrong. In that case I'm not referring to the apple-domain but to literal everything. And yes, everything also contains this pear. I could make the domain-name a part of the identification but that is only possible for my self defined datatypes and not for XQuery's native datatypes. I bit of a complex story but I hope it makes sense to you. Rob -----Oorspronkelijk bericht----- Van: Christian Grün [mailto:christian.gruen@gmail.com] Verzonden: donderdag 30 juli 2015 11:14 Aan: Rob Stapper CC: Eliot Kimber; Florent Georges; BaseX Onderwerp: Re: [basex-talk] question

If I was programming in Java I guess I want Java to return the string: "JLabel" probably combined with the URI of the source were JLabel is defined.

In that case, you would have to work with the reflection features of Java. In BaseX, the equivalent facility is (as you already guessed) the Inspection Module. However, I wouldn't recommend you to follow this path, as you would soon end up doing all kinds of things with string comparisons. What are you actually trying to attain with the string representation of types? --- Dit e-mailbericht is gecontroleerd op virussen met Avast antivirussoftware. https://www.avast.com/antivirus

Typeswitch would indeed help but as I already mentioned it feels a bit, well ... off. It would not be my first choice. Looking at Priscilla's solution I and probably also you can understand why. Yes I also do need string representation for functiontypes: probably a typeswitch will do for map and arrays. I've built me own userdatatypedefinition-module based on closure-functions which already has a function for returning string representations, so that won't be a problem. Since I've a few other modules which would be better off as integrated part of the runtime ( a tracer for instance) I'm thinking about building it into the Basex core in Java. This would ask for mega effort especially since I'm unknown to Java and its develement-environment. Any advise here? Well, lots to do. Rob -----Oorspronkelijk bericht----- Van: Christian Grün [mailto:christian.gruen@gmail.com] Verzonden: donderdag 30 juli 2015 12:52 Aan: Rob Stapper CC: Eliot Kimber; Florent Georges; BaseX Onderwerp: Re: [basex-talk] question

I bit of a complex story but I hope it makes sense to you.

Hm. I still don't see why typeswitch wouldn't help as well, but your use case may indeed be too complex to break it down to a few sentences.. ;) But if you think it cannot be solved with static type checks (i.e., with "typeswitch" and "instance of"), you should probably do what has been motivated by Florent, and write a function that gives you the string representation of types )(see e.g. Priscilla Walmsley's solution for atomic types [1]). Do you also need a string representation of function items? Christian [1] http://www.xqueryfunctions.com/xq/functx_atomic-type.html

- say we have a domain: "pears" containing pears and a domain: "apples" containing apples. - say we define a subset on the "apples"-domain as: "all apples from domain: "apples", except apple: <some_apple_id>, and apple: <some_other_apple_id>" - say we have a pear form the "pears"-domain with id: <some_pear_id>. - if I want to test if the apple-subset contains this pear I must check the domain, i.o.w. type, of the pear against the domain, i.o.w. type, of the apple-subset. If I don't the answer to this question would be: "true", which is wrong. In that case I'm not referring to the apple-domain but to literal everything. And yes, everything also contains this pear.

I could make the domain-name a part of the identification but that is only possible for my self defined datatypes and not for XQuery's native datatypes.

Rob

-----Oorspronkelijk bericht----- Van: Christian Grün [mailto:christian.gruen@gmail.com] Verzonden: donderdag 30 juli 2015 11:14 Aan: Rob Stapper CC: Eliot Kimber; Florent Georges; BaseX Onderwerp: Re: [basex-talk] question

...

If I was programming in Java I guess I want Java to return the string: "JLabel" probably combined with the URI of the source were JLabel is defined.

In that case, you would have to work with the reflection features of Java. In BaseX, the equivalent facility is (as you already guessed) the Inspection Module.

However, I wouldn't recommend you to follow this path, as you would soon end up doing all kinds of things with string comparisons. What are you actually trying to attain with the string representation of types?

--- Dit e-mailbericht is gecontroleerd op virussen met Avast antivirussoftware. https://www.avast.com/antivirus

--- Dit e-mailbericht is gecontroleerd op virussen met Avast antivirussoftware. https://www.avast.com/antivirus