Nabble - Scala - Lounge

Re: Easy serialization

2008-01-11T09:26:33Z

Insitu wrote:

> Hello,
> In the process of writing a piece of software, I ran into the problem
> of serializing or interacting with the user/outside world and am
> looking for a scala(tm) way to deal with that.
>
> I have a class LSystem which I can initialize using "syntactic sugar"
> like:
>
> new LSystem(
> 'A' -> "AB",
> 'B' -> "BB")
>
> I would like to write a small program that read a textual description
> of a l-system and creates a scala object from it. I was thinking of
> at least three possible ways of doing that:
> - write a parser: a bit overkill...
> - lift java serialization (eg. xstream or anything similar): not very
> user friendly
> - use some form dynamic evaluation like in a scala console: I would
> create a l-system object then "evaluate" the rules in its context.

Write a parser or "dynamic evaluation" are for reading stream -> object
How do you plan to write (object -> stream) ?

I did some quick test and xstream work nicely with case class (write and read).
Why did you say "not user friendly" ?

> Is it possible (and easy) to achieve the latter ? Is there another way
> I might be skipping ?
>
> Thanks ,

/davidB

Re: Easy serialization

2008-01-11T06:46:42Z

I'm not sure if that's what you mean...
You should be able to create class LSystem that takes a variable number of LRules in the constructor. LRule class would implement -> method, that sets up the rvalue. You'll also need an implicit def that converts a string to a LRule.

The only thing that (I believe) you won't be able to achieve is to use the 'symbol' notation. As per my understanding, apostrophe is reserved to represent symbols in Scala.

HTH, Stan

On 1/11/08, Insitu <abailly@...> wrote:

Hello,
In the process of writing a piece of software, I ran into the problem
of serializing or interacting with the user/outside world and am
looking for a scala(tm) way to deal with that.

I have a class LSystem which I can initialize using "syntactic sugar"
like:

  new LSystem(
    'A' -> "AB",
    'B' -> "BB")

I would like to write a small program that read a textual description
of a l-system and creates a scala object from it. I was thinking of
at least three possible ways of doing that:
- write a parser: a bit overkill...
- lift java serialization (eg. xstream or anything similar): not very
user friendly
- use some form dynamic evaluation like in a scala console: I would
create a l-system object then "evaluate" the rules in its context.

Is it possible (and easy) to achieve the latter ? Is there another way
I might be skipping ?

Thanks ,
--
Arnaud Bailly, PhD
OQube - Software Engineering
http://www.oqube.com

--
Best wishes,
Stan Klimoff
Grid Dynamics Consulting

Easy serialization

2008-01-11T06:29:34Z

Hello,
In the process of writing a piece of software, I ran into the problem
of serializing or interacting with the user/outside world and am
looking for a scala(tm) way to deal with that.

I have a class LSystem which I can initialize using "syntactic sugar"
like:

new LSystem(
'A' -> "AB",
'B' -> "BB")

I would like to write a small program that read a textual description
of a l-system and creates a scala object from it. I was thinking of
at least three possible ways of doing that:
- write a parser: a bit overkill...
- lift java serialization (eg. xstream or anything similar): not very
user friendly
- use some form dynamic evaluation like in a scala console: I would
create a l-system object then "evaluate" the rules in its context.

Is it possible (and easy) to achieve the latter ? Is there another way
I might be skipping ?

Thanks ,
--
Arnaud Bailly, PhD
OQube - Software Engineering
http://www.oqube.com

End of scala-lounge

2008-01-11T03:43:18Z

Dear scala-lounge subscriber:

scala-lounge has been split into two new lists:

scala-user for newcomers and questions about programming in Scala

scala-debate for more advanced debates, including any form of
language advocacy.

You have been automatically subscribed to both lists. If you wish to
unsubscribe to either list, here's how to do it:

To unsubscribe from scala-user, look here :
http://listes.epfl.ch/cgi-bin/doc_en?liste=scala-user

To unsubscribe from scala-debate, look here :
http://listes.epfl.ch/cgi-bin/doc_en?liste=scala-debate

Re: Overloaded method/function name annoyances

2008-01-11T03:27:59Z

On Jan 10, 2008 1:19 PM, toivo <tom.tad@...> wrote:

>
> Hi,
>
>
> object Annoy1 {
>
> def paymentInv( isa: double, amount: double, ex: double) = isa *
> amount/1000 * ex
>
> def paymentInv = paymentInv( 24, 7000, 0.9)
> }
>
> Error: overloaded method paymentInv needs result type
>
>
> object Annoy1 {
>
> def paymentInv( isa: double, amount: double, ex: double) = isa *
> amount/1000 * ex
>
> val paymentInv = paymentInv( 24, 7000, 0.9)
> }
>
> Error: recursive value paymentInv needs type
>
> Why result type is required?
> And this is not recursive – I am calling different method/function with same
> name but signature is different.
>

Overloaded methods need an explicit result type because
the result type is one of the things that's used to determine
whether a method alternative is applicable or not. There's only so
much one can do. If someone specifies a complete solution to me (in
the language of the JLS) which is not too complicated and which is
implementable, I'd be happy to implement it.

The ``recursive value'' message is a related, but different story --
here, the compiler found that there exist two paymentInv and tried to
determine the correct alternative. In doing that it needed the type of
the value paymentInv, which gave the recursive value error. (and, just
to respond to a possible objection -- the
value paymentInv cannot be ruled out out of hand because
it might be a function value that is applicable to the three arguments).

>
> Overloading (using same method name) works only inside same object/class?

Indeed. The compiler will first look up the members based on the rules
of nesting and import precedence. Then, if there are several eligible
members, overloading resolution is applied.

Cheers

-- Martin

Re: Erlang OTP wire protocol implemented in a Java library

2008-01-10T13:10:57Z

Interoperation with Erlang is a nice property.

Perhaps JGroups or Jini could be used for service discovery but not
for the message transport?

Erlang has some library support to enable group discovery here in a
"process groups" module http://erlang.org/doc/man/pg.html, but the
message transport still uses the standard Erlang protocol.

John

On Jan 10, 2008 11:59 AM, Alex Blewitt <alex.blewitt@...> wrote:

> On Jan 10, 2008 5:34 PM, Bill Venners <bv-jackrabbit@...> wrote:
> > Hi All,
> >
> > For building distributed systems on the JVM, I'd also look at Jini. It
> > has nice tools for discovering other nodes, dealing with partial
> > failure, etc. It's now an Apache project. I'm not sure what the
> > requirements or problems are currently with remote Actors, so I'm not
> > sure if Jini would be a good fit, but it is certainly something I'd
> > recommend looking at.
>
> I think the key point of using the Erlang format is not 'my format's
> better than yours' but because it offers one thing that the others do
> not: seamless interoperation with Erlang.
>
> Alex.
>

--
John D. Heintz
Principal Consultant
New Aspects of Software
http://newaspects.com
http://johnheintz.blogspot.com
Austin, TX
(512) 633-1198

Re: Erlang OTP wire protocol implemented in a Java library

2008-01-10T10:31:48Z

... or even this:

http://jgcs.sourceforge.net/_Media/doa-06-13.pdf

This project is attempting to standardize a group communications API, and (supposedly) already works with JGroups, Appia, and Spread.

Remote actors could even be set up to work as RAIA (redundant array of inexpensive actors), along the lines of RAID and RAIDs (see http://sequoia.continuent.org/HomePage - built using jGCS), where any actor of a suitable type could be chosen from one of many available nodes.

Re: Erlang OTP wire protocol implemented in a Java library

2008-01-10T09:59:45Z

On Jan 10, 2008 5:34 PM, Bill Venners <bv-jackrabbit@...> wrote:
> Hi All,
>
> For building distributed systems on the JVM, I'd also look at Jini. It
> has nice tools for discovering other nodes, dealing with partial
> failure, etc. It's now an Apache project. I'm not sure what the
> requirements or problems are currently with remote Actors, so I'm not
> sure if Jini would be a good fit, but it is certainly something I'd
> recommend looking at.

I think the key point of using the Erlang format is not 'my format's
better than yours' but because it offers one thing that the others do
not: seamless interoperation with Erlang.

Alex.

Re: Erlang OTP wire protocol implemented in a Java library

2008-01-10T09:34:02Z

Hi All,

For building distributed systems on the JVM, I'd also look at Jini. It
has nice tools for discovering other nodes, dealing with partial
failure, etc. It's now an Apache project. I'm not sure what the
requirements or problems are currently with remote Actors, so I'm not
sure if Jini would be a good fit, but it is certainly something I'd
recommend looking at.

Bill

On Jan 10, 2008, at 10:33 AM, Paul J. LaCrosse wrote:

>
>
>
> David Pollak-2 wrote:
>>
>> http://www.theserverside.com/tt/articles/article.tss?l=IntegratingJavaandErlang
>>
>> This might be a good transport layer for remote Actors.
>>
>
> Maybe..., how about JGroups for the transport? (http://
> www.jgroups.org)
>
> The transport stack is very configurable...
> --
> View this message in context: http://www.nabble.com/Erlang-OTP-wire-protocol-implemented-in-a-Java-library-tp14717081p14736481.html
> Sent from the Scala - Lounge mailing list archive at Nabble.com.
>
>

Re: pattern matching on tuples

2008-01-10T08:15:26Z

On Jan 10, 2008 4:55 PM, Carsten Saager <csaager@...> wrote:

>
>
>
>
> On Jan 10, 2008 4:33 PM, Nazar Stasiv <nstasiv@...> wrote:
> > Hello,
> >
> > abstract class Shape
> > case class Circle(radius:Int) extends Shape
> > case class Rectangle(width:Int, height:Int) extends Shape
> >
> > def area(shape:Shape):Double = shape match {
> > case Circle(radius) => 3.14*radius*radius
> > case Rectangle(width, height) => width*height
> > }
> >
> > The code above does pattern matching on of input parameter type
> shape:Shape. In this case it is likely to use tuples to represent data in
> Rectangle and Circle classes.
> > Tuple2[String, Int]("circle", radius)
> > Tuple3[String, Int, Int]("rectangle", width, height)
> > So finally I want to write a code which does pattern matching on tuple.
> Here's how it is done with erlang
> >
> > area({rect,Width,Height})->Width*Height;
> > area({circle,Radius})->3.14*Radius*Radius.
> >
> >
> > Is it possible to have something close in scala? To be specific I mean
> this
> >
> >
> > def area(shape:Tuple):Double = shape match {
> > case ("circle", radius) => 3.14*radius*radius
> > case ("rectangle", width, height) => width*height
> > }
> >
> >
> use overloading - don't match if the compiler can do it for you:
> def area(shape:Tuple2[String, Int]):Double = 3.14*shape._2*shape_2
> def area(shape:Tuple3[String, Int, Int]):Double = shape._2*shape._3
>
> alternatively
>
> def area(shape:Product):Double = shape match {
> case Tuple2("circle", radius:Double) => 3.14*radius*radius
> case Tuple3("rectangle", width:Double, height:Double) => width*height
> }
>
> val circle:Tuple2[String, Int]=("circle", 10)
> val rect:Tuple3[String, Int, Int]=("rectangle", 5, 6)
> Console.println (area(circle))
> Console.println (area(rect))
> Console.println (area(("rectangle",3,2)))
>
> but this is very fragile because of the type you have to define for radius,
> width and height

Tuples can be created and matched with simple parenthesis, like this:

("foo", 42) match { case (s : String, i : Int) => println(s + ", " + i) }

No need to use the TupleX constructors.

/f

Re: pattern matching on tuples

2008-01-10T08:10:42Z

I think, for Scala, your first solution is the correct one. A tuple of 2
elements does not have the same type as a tuple of 3 elements, so you
have to go via Product. What's wrong with the code you posted to start
with, using case classes?

Robin

> On Jan 10, 2008 4:33 PM, Nazar Stasiv <nstasiv@...> wrote:
> abstract class Shape
> case class Circle(radius:Int) extends Shape
> case class Rectangle(width:Int, height:Int) extends Shape
>
> def area(shape:Shape):Double = shape match {
> case Circle(radius) => 3.14*radius*radius
> case Rectangle(width, height) => width*height
> }
>

>
> def area(shape:Product):Double = shape match {
> case Tuple2("circle", radius:Double) => 3.14*radius*radius
> case Tuple3("rectangle", width:Double, height:Double) =>
> width*height
> }

Re: pattern matching on tuples

2008-01-10T07:55:36Z

On Jan 10, 2008 4:33 PM, Nazar Stasiv <nstasiv@...> wrote:

Hello,

abstract class Shape
case class Circle(radius:Int) extends Shape
case class Rectangle(width:Int, height:Int) extends Shape

def area(shape:Shape):Double = shape match {
       case Circle(radius) => 3.14*radius*radius
       case Rectangle(width, height) => width*height
}

The code above does pattern matching on of input parameter type shape:Shape. In this case it is likely to use tuples to represent data in Rectangle and Circle classes.
Tuple2[String, Int]("circle", radius)
Tuple3[String, Int, Int]("rectangle", width, height)
So finally I want to write a code which does pattern matching on tuple. Here's how it is done with erlang

area({rect,Width,Height})->Width*Height;
area({circle,Radius})->3.14*Radius*Radius.

Is it possible to have something close in scala? To be specific I mean this

def area(shape:Tuple):Double = shape match {
       case ("circle", radius) => 3.14*radius*radius
       case ("rectangle", width, height) => width*height
}

use overloading - don't match if the compiler can do it for you:
def area(shape:Tuple2[String, Int]):Double = 3.14*shape._2*shape_2
def area(shape:Tuple3[String, Int, Int]):Double = shape._2*shape._3

alternatively

def area(shape:Product):Double = shape match {
    case Tuple2("circle", radius:Double) => 3.14*radius*radius
    case Tuple3("rectangle", width:Double, height:Double) => width*height
}

    val circle:Tuple2[String, Int]=("circle", 10)
    val rect:Tuple3[String, Int, Int]=("rectangle", 5, 6)
    Console.println (area(circle))
    Console.println (area(rect))
    Console.println (area(("rectangle",3,2)))

but this is very fragile because of the type you have to define for radius, width and height

pattern matching on tuples

2008-01-10T07:33:49Z

Hello,

abstract class Shape
case class Circle(radius:Int) extends Shape
case class Rectangle(width:Int, height:Int) extends Shape

def area(shape:Shape):Double = shape match {
       case Circle(radius) => 3.14*radius*radius
       case Rectangle(width, height) => width*height
}

The code above does pattern matching on of input parameter type shape:Shape. In this case it is likely to use tuples to represent data in Rectangle and Circle classes.
Tuple2[String, Int]("circle", radius)
Tuple3[String, Int, Int]("rectangle", width, height)
So finally I want to write a code which does pattern matching on tuple. Here's how it is done with erlang

area({rect,Width,Height})->Width*Height;
area({circle,Radius})->3.14*Radius*Radius.

Is it possible to have something close in scala? To be specific I mean this

def area(shape:Tuple):Double = shape match {
       case ("circle", radius) => 3.14*radius*radius
       case ("rectangle", width, height) => width*height
}

Thank you.
Nazar

Re: Erlang OTP wire protocol implemented in a Java library

2008-01-10T07:33:47Z

David Pollak-2 wrote:

http://www.theserverside.com/tt/articles/article.tss?l=IntegratingJavaandErlang

This might be a good transport layer for remote Actors.

Maybe..., how about JGroups for the transport? (http://www.jgroups.org)

The transport stack is very configurable...

Re: Re: += and +

2008-01-10T06:09:16Z

"Jamie Webb" <j@...> wrote

>...
> http://lampsvn.epfl.ch/trac/scala/ticket/92
>
> The behaviour is thanks to Scala's precedence rules. The above
> parenthesises as:
>
> (s += "B") + "C"
>
> That is, "B" is appended to s and the result stored back in s, and then
> "C" is appended but that result is discarded.

Now I understand it. In my poor opinion, this is for sure an issue to fix!

Thanks a lot.

Bye

Franco

Re: Re: += and +

2008-01-10T05:25:12Z

On 2008-01-10 14:09:24 Franco Lombardo wrote:
> >> var s = "A"
> >> s += "B" + "C"
> >> println(s)
> > See issue #92.
>
> Please, can you tell me what is issue #92?
> And is there anyone who can explain me the strange behaviour of s +=
> "B" + "C"?

A ticket in the Scala bug tracker:

http://lampsvn.epfl.ch/trac/scala/ticket/92

The behaviour is thanks to Scala's precedence rules. The above
parenthesises as:

(s += "B") + "C"

That is, "B" is appended to s and the result stored back in s, and then
"C" is appended but that result is discarded.

/J

Re: += and +

2008-01-10T05:09:24Z

>> var s = "A"
>> s += "B" + "C"
>> println(s)
> See issue #92.

Please, can you tell me what is issue #92?
And is there anyone who can explain me the strange behaviour of s += "B" +
"C"?
Thanks a lot.

Bye

Franco

Re: += and +

2008-01-10T04:51:56Z

On 2008-01-10 22:16:37 Eric Willigers wrote:
> What do we expect the following to print?
>
> var s = "A"
> s += "B" + "C"
> println(s)
>
> ABC?

See issue #92.

/J

Overloaded method/function name annoyances

2008-01-10T04:19:07Z

Hi,

object Annoy1 {

def paymentInv( isa: double, amount: double, ex: double) = isa * amount/1000 * ex

def paymentInv = paymentInv( 24, 7000, 0.9)
}

Error: overloaded method paymentInv needs result type

object Annoy1 {

def paymentInv( isa: double, amount: double, ex: double) = isa * amount/1000 * ex

val paymentInv = paymentInv( 24, 7000, 0.9)
}

Error: recursive value paymentInv needs type

Why result type is required?
And this is not recursive – I am calling different method/function with same name but signature is different.

package von.lab

object calcEngine {

def paymentInv( isa: double, amount: double, ex: double) = isa * amount/1000 * ex
}

object Annoy2 {

import von.lab.calcEngine._

val paymentInv: double = paymentInv( 24, 7000, 0.9)
}

Error: Annoy2.this.paymentInv of type double does not take parameters

Overloading (using same method name) works only inside same object/class?
Different signature is not enough for compiler?

Am I missing something?

toivo

Re: += and +

2008-01-10T04:11:08Z

On Jan 10, 2008 12:16 PM, Eric Willigers <ewilligers@...> wrote:

What do we expect the following to print?

var s = "A"
s += "B" + "C"
println(s)

ABC?

Do we expect the following to compile?

false + "D"
() + "E"

No, No?

Agree, this is what we see in other (statically typed) languages

It is nice to be able to write println(n + " ...")

I agree that it is nice but this is why we have the strange effects above due to the globally available any2stringad

d - with infix constructors this could be done much cleaner

I'll need to remember to use parens: s += ("B" + "C")

This horrible as val s = "B"+"C" works fine.

Re: Erlang OTP wire protocol implemented in a Java library

2008-01-10T03:35:39Z

David Pollak wrote:
> http://www.theserverside.com/tt/articles/article.tss?l=IntegratingJavaandErlang

Note the conclusion. As Martin has put it, Scala would let him stop
sitting between two chairs.

+= and +

2008-01-10T03:16:37Z

What do we expect the following to print?

var s = "A"
s += "B" + "C"
println(s)

ABC?

Do we expect the following to compile?

false + "D"
() + "E"

No, No?

Actually it prints AB, and they do compile.

Here's what scalac -Xprint:cleanup tells us:-

s = "A";
{
scala.this.Predef.any2stringadd({
s_=(s().+("B"));
scala.runtime.BoxedUnit.UNIT
})
}.+("C");
scala.this.Predef.println(s());
scala.this.Predef.any2stringadd(scala.Boolean.box(false)).+("D");
scala.this.Predef.any2stringadd(scala.runtime.BoxedUnit.UNIT).+("E");

any2stringadd isn't mentioned in the language spec. What is its
motivation? To allow any + string just as string + any is allowed?

It is nice to be able to write println(n + " ...")

I'll need to remember to use parens: s += ("B" + "C")

Re: Type inference with underscore

2008-01-10T02:52:19Z

This a limitation of the implicit parameters. The get bound to the next enclosing parens. Here println() which doesn't provide an argument.

It seems that the Scala compiler purges the redundant parens in println (_) so that this acts as println _, which is OK by the rules of implicit params

Carsten

On Jan 10, 2008 9:33 AM, Michael Pradel <michael@...> wrote:

Hi,

why doesn't the following compile:

object TypeInfer {
def main(args: Array[String]) {
val l = "a" :: "b" :: "c" :: Nil
l.foreach(println(_)) // compiles
l.foreach(println(_ + "xyz")) // doesn't compile
}
}

I expected the type of the second _ to be at least as easy to infer as
that of the first. However, I get:

TypeInfer.scala :5: error: missing parameter type
l.foreach(println(_ + "xyz")) // doesn't compile

Thanks,
Michael

Re: Strange behavior....

2008-01-10T01:10:34Z

Hello (and Happy New Year),
I found the reason for my problem. Here is a skeleton code
illustrating the issues, the reason should be obvious from this code ;-):

trait Oracle {
val outcome : Outcome;
}

abstract class AbstractOracle(val f: Option[File]) extends Oracle {

override val outcome : Outcome = {
f match {
case None => Error
case Some(f) => test(f) // (1)
}
}

def test(f : File) : Outcome;
}

class ConcreteOracle(f1 : Option[File]) extends AbstractOracle(f1) {

var explainFailure : List[Diff] = Nil // (2)

override def test(f : File) : Outcome = {
if(testok(f))
Success
else {
explainFailure = explain(f) // (3)
Failure
}
}
}

The problem is that (2) gets called after (1) because the constructor
code of ConcreteOracle is called after its superclasses'. Hence
explainFailure is correctly computed in (3) and reset to Nil in (2).
Some classical trap with OO constructor dispatch...

Everything works fine when I write :

abstract class AbstractOracle(val f: Option[File]) extends Oracle {

override lazy val outcome : Outcome = {

Then outcome is called only when requested, which does occur after the
constructor sequence has finished.

BTW, lazy values does not seem to be memoized (according to the
generated bytecode at least) or maybe I missed something. Is there a
deep reason for this ? I think that Haskell's (and maybe OCaml's) lazy
values are memoized which leads to better performances.

Best regards,

--
Arnaud Bailly, PhD
OQube - Software Engineering
http://www.oqube.com

Type inference with underscore

2008-01-10T00:33:07Z

Hi,

why doesn't the following compile:

object TypeInfer {
def main(args: Array[String]) {
val l = "a" :: "b" :: "c" :: Nil
l.foreach(println(_)) // compiles
l.foreach(println(_ + "xyz")) // doesn't compile
}
}

I expected the type of the second _ to be at least as easy to infer as
that of the first. However, I get:

TypeInfer.scala:5: error: missing parameter type
l.foreach(println(_ + "xyz")) // doesn't compile

Thanks,
Michael

Blog posting for Scala newbies

2008-01-09T13:07:52Z

http://www.codecommit.com/blog/scala/scala-for-java-refugees-part-1

Re: scala erb equivalent?

2008-01-09T11:24:22Z

Second, you could do something like:

def showIf(test: Boolean)(toShow: => NodeSeq) = if (test) toShow else Text("")

very neat! I still dont know the language enough to realize things like that are doable. thanks.

I think Jim's solution is pretty cool too, and generally useful for simple string interpolation.. maybe something like that would be good to add to scalax/scalaz ?

davidB brings up a good point-- I hadn't really looked at many of the templating solutions from Java yet.. I'll take a look at those before trying to roll my own erb-clone (which I might do anyway for fun).

thanks for all the replies!

Re: scala erb equivalent?

2008-01-09T11:04:11Z

Hi,

Scala can use all the templates lib available in the javaworld like velocity, freemarker, stringtemplate,...
Or simple like the code provide by Jim cf Substitutor from commons-lang.

/davidB

Jim Miller wrote:

> I use the following function that goes through an imput string and calls
> a function for every occurrence of #{name}, passing 'name' as the
> argument to the function.
>
> import java.util.regex.{Pattern,Matcher}
>
> def populate (template: String, map : (String) => String ) : String = {
>
> val p = Pattern.compile("#\\{(.+?)\\}");
> val m = p.matcher(template);
> val sb = new StringBuffer();
>
> while (m.find()) {
> m.appendReplacement(sb, map( m.group(1) ) );
> }
>
> m.appendTail(sb);
>
> return sb.toString()
> }
>
> Then in practice I do something like:
>
> populate(
> """XDR.writeInt32(d, #{name}.size() );
> for ( #{type} v : #{name} ) { XDR.write( d, v ); }
> """,
> tag => tag match {
> case "name" => "foo"
> case "type" => "bar"
> })
>
> I do a LOT of code generation with Scala and this function works out
> very well. The nice thing is that I can do anything I want in the case
> clauses as long as the result is a string for substituting in the template.
>
> On 1/9/08, *Maurice Codik* <mcodik@... <mailto:mcodik@...>>
> wrote:
>
> Hey all,
>
> I'm looking for a library similar to Ruby's ERB for scala: something
> that I can use to embed scala code in ordinary text, and then have
> that translated into a scala class I can call into. I know I can get
> close to this by using { expr } in xml tags, but I'd like something
> I can use outside of xml documents if I need to.
>
> Does something like this exist already? My google-fu is failing me..
>
> Thanks!
>
> Maurice
>
>
>

Re: scala erb equivalent?

2008-01-09T10:58:04Z

Maurice,

Maurice Codik wrote:

A few reasons:

- not all documents are xml.

- I've run into a few issues with the built in xml support that make it annoying for use as a templating language for xml. for example, with simple conditionals:

object xml {
def main(a:Array[String]) {
    val data = <testing>
                {
                  if (true) {
                    <huh>huh?!</huh>
                  }
               }
               </testing>
    Console.println(data)
}
}

only prints: <testing></testing>. I can work around this by adding a dummy else block to that if, but that can get tedious.

First, this is good practice for remembering that everything in Scala returns a value, but sometimes the value is of type Unit.

Second, you could do something like:

def showIf(test: Boolean)(toShow: => NodeSeq) = if (test) toShow else Text("")

Then you could write showIf(true){<huh>huh?!</huh>}

there are also some slightly annoying things, like how it turns self closing tags to normal tags: <img src="test.jpg" /> => <img ..></img>

There's an XML to String to UTF-8 encoded byte array facility in lift that "does the right thing"(tm) for XML -> XHTMLish stuff that even IE 6 is comfortable with.

On Jan 9, 2008 9:57 AM, David Pollak < feeder.of.the.bears@...> wrote:

Short answer: nothing but the built-in XML support.

Why is this failing you?

On 1/9/08, Maurice Codik <mcodik@...> wrote:
Hey all,

I'm looking for a library similar to Ruby's ERB for scala: something that I can use to embed scala code in ordinary text, and then have that translated into a scala class I can call into. I know I can get close to this by using { expr } in xml tags, but I'd like something I can use outside of xml documents if I need to.

Does something like this exist already? My google-fu is failing me..

Thanks!

Maurice

--
lift, the secure, simple, powerful web framework http://liftweb.net
Collaborative Task Management http://much4.us

Re: scala erb equivalent?

2008-01-09T10:29:48Z

A few reasons:

- not all documents are xml.

- I've run into a few issues with the built in xml support that make it annoying for use as a templating language for xml. for example, with simple conditionals:

object xml {
def main(a:Array[String]) {
    val data = <testing>
                {
                  if (true) {
                    <huh>huh?!</huh>
                  }
               }
               </testing>
    Console.println(data)
}
}

only prints: <testing></testing>. I can work around this by adding a dummy else block to that if, but that can get tedious.

there are also some slightly annoying things, like how it turns self closing tags to normal tags: <img src="test.jpg" /> => <img ..></img>

On Jan 9, 2008 9:57 AM, David Pollak < feeder.of.the.bears@...> wrote:

Short answer: nothing but the built-in XML support.

Why is this failing you?

On 1/9/08, Maurice Codik <mcodik@...> wrote:
Hey all,

I'm looking for a library similar to Ruby's ERB for scala: something that I can use to embed scala code in ordinary text, and then have that translated into a scala class I can call into. I know I can get close to this by using { expr } in xml tags, but I'd like something I can use outside of xml documents if I need to.

Does something like this exist already? My google-fu is failing me..

Thanks!

Maurice

--
lift, the secure, simple, powerful web framework http://liftweb.net
Collaborative Task Management http://much4.us

Re: scala erb equivalent?

2008-01-09T10:25:35Z

I use the following function that goes through an imput string and calls a function for every occurrence of #{name}, passing 'name' as the argument to the function.

import java.util.regex.{Pattern,Matcher}

def populate (template: String, map : (String) => String ) : String = {

val p = Pattern.compile("#\\{(.+?)\\}");
val m = p.matcher(template);
val sb = new StringBuffer();

while (m.find()) {
    m.appendReplacement(sb, map( m.group(1) ) );
}

m.appendTail(sb);

return sb.toString()
}

Then in practice I do something like:

    populate(
      """XDR.writeInt32(d, #{name}.size() );
          for ( #{type} v : #{name} ) { XDR.write( d, v ); }
      """,
      tag => tag match {
   case "name" => "foo"
        case "type" => "bar"
      })

I do a LOT of code generation with Scala and this function works out very well. The nice thing is that I can do anything I want in the case clauses as long as the result is a string for substituting in the template.

On 1/9/08, Maurice Codik <mcodik@...> wrote:

Hey all,

I'm looking for a library similar to Ruby's ERB for scala: something that I can use to embed scala code in ordinary text, and then have that translated into a scala class I can call into. I know I can get close to this by using { expr } in xml tags, but I'd like something I can use outside of xml documents if I need to.

Does something like this exist already? My google-fu is failing me..

Thanks!

Maurice

Re: By-name parameter oddities

2008-01-09T10:15:57Z

martin odersky wrote:

> On Jan 9, 2008 1:19 AM, Blair Zajac <blair@...> wrote:
>> 1) I'm guessing the following should be a compile error, given that you cannot
>> return a by-name type:
>>
>> abstract class Foo {
>> def identity[T](x : => T) : ( => T)
>> }
>>
> Right. That should not be permitted.
>
> -- Martin
>

I've opened a ticket:

http://lampsvn.epfl.ch/trac/scala/ticket/351

Blair

Re: scala erb equivalent?

2008-01-09T09:57:35Z

Short answer: nothing but the built-in XML support.

Why is this failing you?

On 1/9/08, Maurice Codik <mcodik@...> wrote:

Hey all,

I'm looking for a library similar to Ruby's ERB for scala: something that I can use to embed scala code in ordinary text, and then have that translated into a scala class I can call into. I know I can get close to this by using { expr } in xml tags, but I'd like something I can use outside of xml documents if I need to.

Does something like this exist already? My google-fu is failing me..

Thanks!

Maurice

--
lift, the secure, simple, powerful web framework http://liftweb.net
Collaborative Task Management http://much4.us

scala erb equivalent?

2008-01-09T09:37:07Z

Hey all,

I'm looking for a library similar to Ruby's ERB for scala: something that I can use to embed scala code in ordinary text, and then have that translated into a scala class I can call into. I know I can get close to this by using { expr } in xml tags, but I'd like something I can use outside of xml documents if I need to.

Does something like this exist already? My google-fu is failing me..

Thanks!

Maurice

Erlang OTP wire protocol implemented in a Java library

2008-01-09T08:59:41Z

http://www.theserverside.com/tt/articles/article.tss?l=IntegratingJavaandErlang

This might be a good transport layer for remote Actors.