Thoughts about RichObjects, Implicit convertions and performance
|
View:
New views
4 Messages
—
Rating Filter:
Alert me
|
| < Prev | 1 - 2 | Next > |
 |
Re: Thoughts about RichObjects, Implicit convertions and performance
by Ben Hutchison-2
::
Rate this Message:
As David Hall mentioned, a similar idea has been raised before (by David
MacIver, IIRC) and received a fair amount of support, including me. A search of Nabble and my archives hasn't located it, unfortunately, but it was worth reading. Can anyone post a ref to it?? As I recall, the prev. proposal required little syntactic change to effect. -Ben Vladimir Kirichenko wrote: > One of the typical uses of implicit conversions is to declare > "additional" methods for the existing class. Compiler substitutes > reference to object with invocation of implicit function that typically > instantiates some kind of RichObject or anonymous object with additional > functions. For example > > var i = 5; > i = i max 6; > i = i max 7; > > will be compiled to 2 instantiations of RichInt. The use of anonymous > declaration in implicit will cause big performance degradation. > > Also in these cases usually impossible to reuse created instance. > > The idea is to create some kind of "implicit traits" for this kind of > situations that would not cause performance issues and also eliminates > the need of implicit def declarations: > > implicit[Int] trait RichInt { > def max(x: Int) = this > x > def min(x: Int) = this < x > } > > and the > > i = i max 6; > > will actually be compiled to "static" method invocation > > i = RichInt.max(i, 6); > > The performance difference is significant > > object XT { > > implicit def int2x(i: Int) = new { > def sum(x: Int) = i + x > } > > def sum(i:Int, x:Int) = i + x; > > def main(args : Array[String]) = { > > //simple > val iterations = 10000000 > var start = System.currentTimeMillis > > var i=0; > var x=0; > > while(i < iterations) { > x = x sum i sum i; > i = i + 1; > } > > println(x + " in " + (System.currentTimeMillis - start) ); > > //fast > start = System.currentTimeMillis > > i=0; > x=0; > > while(i < iterations) { > x = sum(sum(x, i),i); > i = i + 1; > } > > println(x + " in " + (System.currentTimeMillis - start) ); > > } > } > ============= > 266447232 in 7557 > 266447232 in 20 > > > > -- *Ben Hutchison Senior Developer * Level 2 476 St Kilda Road Melbourne VIC 3004 T 613 8807 5252 | F 613 8807 5203 | M 0423 879 534 | www.ibsglobalweb.com <http://www.ibsglobalweb.com/> |
|
|
Re: Thoughts about RichObjects, Implicit convertions and performance
by David Hall-15
::
Rate this Message:
On Tue, Mar 24, 2009 at 11:50 PM, Ben Hutchison <benh@...> wrote:
> As David Hall mentioned, a similar idea has been raised before (by David > MacIver, IIRC) and received a fair amount of support, including me. > > A search of Nabble and my archives hasn't located it, unfortunately, but it > was worth reading. Can anyone post a ref to it?? As I recall, the prev. > proposal required little syntactic change to effect. Apparently it was Jorge, and apparently the syntax he proposed was the same one I mentioned above... I could have sworn it was different... http://www.nabble.com/Implicit-classes-(or-encouraging-people-to-pimp-hygienically)-p18192147.html The main change I would make to his proposal is to allow implicit classes to actually perform the conversion when necessary, but to elide it out whenever possible. -- David > > -Ben > > Vladimir Kirichenko wrote: >> >> One of the typical uses of implicit conversions is to declare >> "additional" methods for the existing class. Compiler substitutes >> reference to object with invocation of implicit function that typically >> instantiates some kind of RichObject or anonymous object with additional >> functions. For example >> >> var i = 5; >> i = i max 6; >> i = i max 7; >> >> will be compiled to 2 instantiations of RichInt. The use of anonymous >> declaration in implicit will cause big performance degradation. >> >> Also in these cases usually impossible to reuse created instance. >> >> The idea is to create some kind of "implicit traits" for this kind of >> situations that would not cause performance issues and also eliminates >> the need of implicit def declarations: >> >> implicit[Int] trait RichInt { >> def max(x: Int) = this > x >> def min(x: Int) = this < x >> } >> >> and the >> >> i = i max 6; >> >> will actually be compiled to "static" method invocation >> >> i = RichInt.max(i, 6); >> >> The performance difference is significant >> >> object XT { >> >> implicit def int2x(i: Int) = new { >> def sum(x: Int) = i + x >> } >> >> def sum(i:Int, x:Int) = i + x; >> >> def main(args : Array[String]) = { >> >> //simple >> val iterations = 10000000 >> var start = System.currentTimeMillis >> >> var i=0; >> var x=0; >> >> while(i < iterations) { >> x = x sum i sum i; >> i = i + 1; >> } >> >> println(x + " in " + (System.currentTimeMillis - start) ); >> >> //fast >> start = System.currentTimeMillis >> >> i=0; >> x=0; >> >> while(i < iterations) { >> x = sum(sum(x, i),i); >> i = i + 1; >> } >> >> println(x + " in " + (System.currentTimeMillis - start) ); >> >> } >> } >> ============= >> 266447232 in 7557 >> 266447232 in 20 >> >> >> >> > > > -- > > > > *Ben Hutchison > Senior Developer > * Level 2 476 St Kilda Road Melbourne VIC 3004 > T 613 8807 5252 | F 613 8807 5203 | M 0423 879 534 | www.ibsglobalweb.com > <http://www.ibsglobalweb.com/> > > > |
|
|
Re: Thoughts about RichObjects, Implicit convertions and performance
by Jorge Ortiz-3
::
Rate this Message:
I'd agree with that change, as long as there is an annotation (a la @tailrec) that ensures a conversion can be elided or emits a compiler error.
--j On Wed, Mar 25, 2009 at 10:59 AM, David Hall <dlwh@...> wrote:
|
|
|
Re: Thoughts about RichObjects, Implicit convertions and performance
by Martin Odersky
::
Rate this Message:
I'm late to the debate...
I think the idea of allowing implicit for classes has a lot of merit. I am not sure we need to spec the generation of static methods, though. I see this rather as a useful a optimization, if the normal scheme is too slow. Cheers -- Martin On Wed, Mar 25, 2009 at 7:59 PM, David Hall <dlwh@...> wrote: > On Tue, Mar 24, 2009 at 11:50 PM, Ben Hutchison <benh@...> wrote: >> As David Hall mentioned, a similar idea has been raised before (by David >> MacIver, IIRC) and received a fair amount of support, including me. >> >> A search of Nabble and my archives hasn't located it, unfortunately, but it >> was worth reading. Can anyone post a ref to it?? As I recall, the prev. >> proposal required little syntactic change to effect. > > Apparently it was Jorge, and apparently the lot syntax he proposed was the > same one I mentioned above... I could have sworn it was different... > > http://www.nabble.com/Implicit-classes-(or-encouraging-people-to-pimp-hygienically)-p18192147.html > > The main change I would make to his proposal is to allow implicit > classes to actually perform the conversion when necessary, but to > elide it out whenever possible. > > -- David > >> >> -Ben >> >> Vladimir Kirichenko wrote: >>> >>> One of the typical uses of implicit conversions is to declare >>> "additional" methods for the existing class. Compiler substitutes >>> reference to object with invocation of implicit function that typically >>> instantiates some kind of RichObject or anonymous object with additional >>> functions. For example >>> >>> var i = 5; >>> i = i max 6; >>> i = i max 7; >>> >>> will be compiled to 2 instantiations of RichInt. The use of anonymous >>> declaration in implicit will cause big performance degradation. >>> >>> Also in these cases usually impossible to reuse created instance. >>> >>> The idea is to create some kind of "implicit traits" for this kind of >>> situations that would not cause performance issues and also eliminates >>> the need of implicit def declarations: >>> >>> implicit[Int] trait RichInt { >>> def max(x: Int) = this > x >>> def min(x: Int) = this < x >>> } >>> >>> and the >>> >>> i = i max 6; >>> >>> will actually be compiled to "static" method invocation >>> >>> i = RichInt.max(i, 6); >>> >>> The performance difference is significant >>> >>> object XT { >>> >>> implicit def int2x(i: Int) = new { >>> def sum(x: Int) = i + x >>> } >>> >>> def sum(i:Int, x:Int) = i + x; >>> >>> def main(args : Array[String]) = { >>> >>> //simple >>> val iterations = 10000000 >>> var start = System.currentTimeMillis >>> >>> var i=0; >>> var x=0; >>> >>> while(i < iterations) { >>> x = x sum i sum i; >>> i = i + 1; >>> } >>> >>> println(x + " in " + (System.currentTimeMillis - start) ); >>> >>> //fast >>> start = System.currentTimeMillis >>> >>> i=0; >>> x=0; >>> >>> while(i < iterations) { >>> x = sum(sum(x, i),i); >>> i = i + 1; >>> } >>> >>> println(x + " in " + (System.currentTimeMillis - start) ); >>> >>> } >>> } >>> ============= >>> 266447232 in 7557 >>> 266447232 in 20 >>> >>> >>> >>> >> >> >> -- >> >> >> >> *Ben Hutchison >> Senior Developer >> * Level 2 476 St Kilda Road Melbourne VIC 3004 >> T 613 8807 5252 | F 613 8807 5203 | M 0423 879 534 | www.ibsglobalweb.com >> <http://www.ibsglobalweb.com/> >> >> >> > |
| < Prev | 1 - 2 | Next > |
| Free embeddable forum powered by Nabble | Forum Help |
