Re: Announce: ErsatzLisp (Java PicoLisp)
On Fri, Dec 09, 2011 at 07:56:52AM +0100, Alexander Burger wrote: > > nice to have a link to this kind of Ersatz documentation from the > > wiki. > > Good idea. I can prepare a first version. Here is my first try: http://picolisp.com/5000/!wiki?ErsatzReflection It can also be found via the "Documentation" menu. Please feel free to improve it! Cheers, - Alex -- UNSUBSCRIBE: mailto:picolisp@software-lab.de?subject=Unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Jon, > Is the "update" below what's available on Ersatz documentation? I > couldn't find this info anywhere else. I think it would have been Yes, I think that's all :( > nice to have a link to this kind of Ersatz documentation from the > wiki. Good idea. I can prepare a first version. You might then add your experiences, if you like. Cheers, - Alex -- UNSUBSCRIBE: mailto:picolisp@software-lab.de?subject=Unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Alex, Is the "update" below what's available on Ersatz documentation? I couldn't find this info anywhere else. I think it would have been nice to have a link to this kind of Ersatz documentation from the wiki. /Jon On 11/4/10 7:53 PM, Alexander Burger wrote: Hi all, here is an update. Besides the basic I/O system and half-baked network functions (until now only 'connect', 'accept' and a partial 'port'), we have now a set of functions interfacing to the Java Reflection API. I'd like to describe them shortly, and hear your opinions. You can try it if you like if you download the current testing release. The central function is called 'java'. It comes in three forms: (java 'cls 'T 'any ..) -> obj This is a constructor call, returning a Java object. 'cls' is the name of a Java class, like "java.lang.StringBuilder". (java 'cls 'msg 'any ..) -> obj This calls a static method 'msg' for a class 'cls'. (java 'obj 'msg 'any ..) -> obj This calls a dynamic method 'msg' for an object 'obj' Example: : (setq Sb (java "java.lang.StringBuilder" T "abc")) -> $StringBuilder : (java Sb "append" (char: 44)) -> $StringBuilder : (java Sb "append" 123) -> $StringBuilder : (java Sb "toString") -> $String : (data @) -> "abc,123" The function 'data' in the last line converts back from Java objects to PicoLisp data. Then we have 'public', it supports two forms: (public 'obj 'any) -> obj Returns the value of a public field 'any' in object 'obj' (public 'cls 'any) -> obj Returns the value of a public field 'any' in class 'cls' Example: : (public "java.lang.System" "err") -> $PrintStream : (java @ "println" "Hello world") Hello world -> NIL Finally, we have a set of type conversion functions, to produce Java objects from Lisp data: (byte: 'num|sym) -> obj (char: 'num|sym) -> obj (int: 'num) -> obj (long: 'num) -> obj (big: 'num) -> obj We saw the usage of 'char:' in the StringBuilder example above. So the 'any' arguments to the 'java' function can either be - 'T' or 'NIL', then the type is boolean - A number, then it must fit in 32 bits and will be passed as 'int' - A normal symbol, then it will be passed as 'String' - A Java object, as returned by 'java' or one of the 'xxx:' conversion functions. - A list, then all elements must be of the same type and will be passed as 'Array' I hope this reflection interface is a good compromise between simplicity and usefulness. Cheers, - Alex -- UNSUBSCRIBE: mailto:picolisp@software-lab.de?subject=Unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Tomas, > But my original point was, can you use Class instances directly in > ErsatzLisp? E.g. something like No, I didn't want to go that far. I just pass the class names. After all, the dynamic lookup of the class during reflection is neglectible compared to the rest of the overhead. You just get more adminstrative overhead of you keep the classes around when you just need to handle the names. But it could be easily added, of course. Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Alex, >>(java "javax.swing.JFrame" T "Animation") >> >> or >> >>(interface "java.awt.event.MouseListener" ... >> >> In wl, there is no need for this because I can do this at read-time, >> like the above example: >> >>`(jclass "java.lang.System") >> >> Also, I can do it only once, see the 'import' function and how it is >> used in swing.l and swt.l. > > This is also the same in ErsatzLisp. 'interface' returns a Proxy object > (as 'java' returns an arbitrary Java object) which can be kept in a > variable etc. Ok. But my original point was, can you use Class instances directly in ErsatzLisp? E.g. something like (interface MouseListener ... where (val 'MouseListener) instanceof java.lang.Class for "java.awt.event.MouseListener" (in other words java.awt.event.MouseListener.class) Cheers, Tomas -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Tomas, > > I cannot see, though, whether you also handle a mapping of Lisp lists > > from/to Java arrays, which doubles the code in the case of ErsatzLisp. > > I think there are two functions: 'jv2l' and 'jvector'. I see. I cannot find out how it works, however. I tried: : (setq S (jnew (jclass 'java.lang.String) "abcde")) -> "abcde" : (S 'getBytes) -> [...@1027b4d] How do I retrieve the actual bytes then? I tried 'jv2l' and 'jvector' but get an error "No Cons.obj". For example, in ErsatzLisp I do this as : (setq S (java "java.lang.String" T "abcde")) -> $String : (java (java S "getBytes")) -> (97 98 99 100 101) You also do the opposite (construct a string out of single bytes) : (java "java.lang.String" T (mapcar byte: (100 101 102))) -> $String : (java @) -> "def" Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Tomas, > > BTW, how can I access existing objects (like "java.lang.System.err") > > in wl? I can only find 'jnew', which creates new objects. > > Like this: > >(jfield `(jclass "java.lang.System") 'err) I see. Thanks! > Basically, because Java objects are > first class citizens in the wl interpreter, I need very little special The same is the case in ErsatzLisp. > > possibly want to use ErsatzLisp in real applications (e.g. Android > > apps) if I find the time. > > Well, to be pedantic in regards to speed optimisations, whenever you do > something class specific you perform a class lookup based on the class Right. But such calls happen infrequent compared to the execution of the language itself. To make it practically usable, I want the raw Lisp code to run as fast as possible. There I'm not satisfied even with ErsatzLisp, and wl is still another factor slower. >(java "javax.swing.JFrame" T "Animation") > > or > >(interface "java.awt.event.MouseListener" ... > > In wl, there is no need for this because I can do this at read-time, > like the above example: > >`(jclass "java.lang.System") > > Also, I can do it only once, see the 'import' function and how it is > used in swing.l and swt.l. This is also the same in ErsatzLisp. 'interface' returns a Proxy object (as 'java' returns an arbitrary Java object) which can be kept in a variable etc. Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Alex, > However, I see that in the function 'jarg' you are doing the same (check > for all kinds of types). yes, it happens in Java methods: 'jarg' and 'isInstance'. These are necessary to automatically handle primitive types and I think it is impossible to reduce it further while preserving the functionality. > I cannot see, though, whether you also handle a mapping of Lisp lists > from/to Java arrays, which doubles the code in the case of ErsatzLisp. I think there are two functions: 'jv2l' and 'jvector'. Cheers, Tomas -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Alex, > BTW, how can I access existing objects (like "java.lang.System.err") > in wl? I can only find 'jnew', which creates new objects. Like this: (jfield `(jclass "java.lang.System") 'err) There are some useful functions defined in java.wl file, implemented in picolisp instead of in native java. Basically, because Java objects are first class citizens in the wl interpreter, I need very little special purpose native functions and the rest can be build on top of that in picolisp. > It is just that I've put much more stress on execution speed, as I > possibly want to use ErsatzLisp in real applications (e.g. Android > apps) if I find the time. Well, to be pedantic in regards to speed optimisations, whenever you do something class specific you perform a class lookup based on the class name passed in: e.g. (java "javax.swing.JFrame" T "Animation") or (interface "java.awt.event.MouseListener" ... In wl, there is no need for this because I can do this at read-time, like the above example: `(jclass "java.lang.System") Also, I can do it only once, see the 'import' function and how it is used in swing.l and swt.l. Cheers, Tomas -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Tomas,
> >> > look at lines 99 through 142 of "ersatz/fun.src" (actual version),
> >> > you see a lot of tedious type ('instanceof') checks. And they don't
> >> > even cover all possible cases! The same would be required for the
> >> > opposite direction. Or does anybody know a better way?
> >
> > I say that the above code is a mess because it still has to check for
> > each possible type and take the appropriate action. As you can see, it
> > covers bytes, characters, integers, longs, doubles, bigIntegers,
> > strings, objects, byte arrays, char arrays, int arrays, long arrays and
> > double arrays. Other types, like short integers, or string arrays etc.
> > are not covered (and I hope they are not such important). BTW, I just
> > notice that booleans (and boolean arrays) are also missig; booleans
> > might be needed.
> >
> > I cannot see a generic way to do that.
>
> I haven't looked at your code, but you might have a look at wl
> interpreter we discussed some time ago. The java ffi was quite complete
Yes, thank you!
However, I see that in the function 'jarg' you are doing the same (check
for all kinds of types).
I cannot see, though, whether you also handle a mapping of Lisp lists
from/to Java arrays, which doubles the code in the case of ErsatzLisp.
> and quite generic and more importantly, completely automatic, althought
> not the fastest due to relying on reflection completely. There is code
> for finding the right method and constructor with matching prototypes
In that sense, the same happens in ErsatzLisp.
It is just that I've put much more stress on execution speed, as I
possibly want to use ErsatzLisp in real applications (e.g. Android apps)
if I find the time.
BTW, how can I access existing objects (like "java.lang.System.err") in
wl? I can only find 'jnew', which creates new objects.
Cheers,
- Alex
--
UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Alex,
>> > look at lines 99 through 142 of "ersatz/fun.src" (actual version),
>> > you see a lot of tedious type ('instanceof') checks. And they don't
>> > even cover all possible cases! The same would be required for the
>> > opposite direction. Or does anybody know a better way?
>
> I say that the above code is a mess because it still has to check for
> each possible type and take the appropriate action. As you can see, it
> covers bytes, characters, integers, longs, doubles, bigIntegers,
> strings, objects, byte arrays, char arrays, int arrays, long arrays and
> double arrays. Other types, like short integers, or string arrays etc.
> are not covered (and I hope they are not such important). BTW, I just
> notice that booleans (and boolean arrays) are also missig; booleans
> might be needed.
>
> I cannot see a generic way to do that.
I haven't looked at your code, but you might have a look at wl
interpreter we discussed some time ago. The java ffi was quite complete
and quite generic and more importantly, completely automatic, althought
not the fastest due to relying on reflection completely. There is code
for finding the right method and constructor with matching prototypes
and something similar for fields to get and set them IIRC.
Cheers,
Tomas
--
UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
On Thu, Nov 18, 2010 at 04:11:24PM +0100, Jakob Eriksson wrote:
> > look at lines 99 through 142 of "ersatz/fun.src" (actual version), you
> > see a lot of tedious type ('instanceof') checks. And they don't even
> > cover all possible cases! The same would be required for the opposite
> > direction. Or does anybody know a better way?
>
> Is this related to introspection?
> http://stackoverflow.com/questions/206/java-introspection-and-reflection
Yes, I'm using the reflection API to find out classes and fields.
I say that the above code is a mess because it still has to check for
each possible type and take the appropriate action. As you can see, it
covers bytes, characters, integers, longs, doubles, bigIntegers,
strings, objects, byte arrays, char arrays, int arrays, long arrays and
double arrays. Other types, like short integers, or string arrays etc.
are not covered (and I hope they are not such important). BTW, I just
notice that booleans (and boolean arrays) are also missig; booleans
might be needed.
I cannot see a generic way to do that.
Cheers,
- Alex
--
UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
On 11/17/2010 06:13 PM, Alexander Burger wrote:
>
> But more important is that this would require quite a mess of code,
> analog to what is there for getting the values of public fields. If you
> look at lines 99 through 142 of "ersatz/fun.src" (actual version), you
> see a lot of tedious type ('instanceof') checks. And they don't even
> cover all possible cases! The same would be required for the opposite
> direction. Or does anybody know a better way?
Is this related to introspection?
http://stackoverflow.com/questions/206/java-introspection-and-reflection
// Jakob
--
UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Alex, I'll try to find an example to test it. ;-) /Jon On 18-11-10 09:50 , Alexander Burger wrote: Hi Jon, (public 'obj 'any) -> obj Returns the value of a public field 'any' in object 'obj' (public 'cls 'any) -> obj Returns the value of a public field 'any' in class 'cls' I would suggest that this 'public' function should allow multiple any's, so that you can do a chained get-value. ;-) This is a good idea, as it is along the line of the 'get' family of functions. I added this feature, and released a new version. Do you know of an example to test it? Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Jon, > >(public 'obj 'any) -> obj > > Returns the value of a public field 'any' in object 'obj' > > > >(public 'cls 'any) -> obj > > Returns the value of a public field 'any' in class 'cls' > I would suggest that this 'public' function should allow multiple > any's, so that you can do a chained get-value. ;-) This is a good idea, as it is along the line of the 'get' family of functions. I added this feature, and released a new version. Do you know of an example to test it? Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi again, On 04-11-10 19:53 , Alexander Burger wrote: Hi all, here is an update. Besides the basic I/O system and half-baked network functions (until now only 'connect', 'accept' and a partial 'port'), we have now a set of functions interfacing to the Java Reflection API. I'd like to describe them shortly, and hear your opinions. You can try it if you like if you download the current testing release. .. Then we have 'public', it supports two forms: (public 'obj 'any) -> obj Returns the value of a public field 'any' in object 'obj' (public 'cls 'any) -> obj Returns the value of a public field 'any' in class 'cls' I would suggest that this 'public' function should allow multiple any's, so that you can do a chained get-value. ;-) /Jon -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
I've just tried it on my Windows 7 (64bit) with JRE 1.6.0_22 (32bit) It works fine, the only thing is java is not in my PATH, so I needed to write full path to java.exe :) "C:\Program Files (x86)\Java\jre6\bin\java.exe" -DPID=42 -jar picolisp.jar lib.l On Wed, Nov 17, 2010 at 07:58:20PM +0100, Alexander Burger wrote: java -DPID=42 -jar picolisp.jar -"on *Dbg" lib.l Further simplified now: The (on *Dbg) moved to "lib.l". Thus java -DPID=42 -jar picolisp.jar lib.l is enough now. Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
On Wed, Nov 17, 2010 at 07:58:20PM +0100, Alexander Burger wrote: >java -DPID=42 -jar picolisp.jar -"on *Dbg" lib.l Further simplified now: The (on *Dbg) moved to "lib.l". Thus java -DPID=42 -jar picolisp.jar lib.l is enough now. Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
On Wed, Nov 17, 2010 at 06:54:52PM +0100, Alexander Burger wrote: > Anyone care to try it under another OS, e.g. Windoofs? > > This would probably require making a batch file "ersatz/picolisp.bat", > along the line of "ersatz/picolisp". The PID is probably not available > there, so some arbitrary number might be used. Josef Bartl just tried it for me. The line java -DPID=42 -jar picolisp.jar -"on *Dbg" lib.l works just fine under Windows. To make a proper batch file, I think something like java -DPID=42 -jar picolisp.jar -"on *Dbg" lib.l %1 %2 %3 %4 %5 %6 should be used, right? Or is there some better way, corresponding to the "$@" argument expansion in bash? The dummy PID (here 42) should better be replaced by some number unique at each invocation, but I have no idea. Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi all, now I simplified the file structure of Ersatz PicoLisp a little, to make it better usable as a stand-alone system. The start-up script 'erl' (as it was described in "ersatz/README" and "INSTALL") doesn't exist any longer, and it now can (or should) be started directly as $ ersatz/picolisp : A minimal package is also available, indepentent from the standard Picolisp release, under http://software-lab.de/ersatz.tgz It contains the README, but no sources or documentation. There are only three runtime files: The "picolisp" start-up script, the JAR file and a combined "lib.l". Anyone care to try it under another OS, e.g. Windoofs? This would probably require making a batch file "ersatz/picolisp.bat", along the line of "ersatz/picolisp". The PID is probably not available there, so some arbitrary number might be used. Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Jon,
> In your description below I can see no way to
> - set a value of a public field of an object, or
> - set a value of a public field of a class
You are right.
> Is this something that you may add later?
Hmm, I would actually prefer not to add it. For one thing, the political
correct way in Java is using set() methods for that.
But more important is that this would require quite a mess of code,
analog to what is there for getting the values of public fields. If you
look at lines 99 through 142 of "ersatz/fun.src" (actual version), you
see a lot of tedious type ('instanceof') checks. And they don't even
cover all possible cases! The same would be required for the opposite
direction. Or does anybody know a better way?
Cheers,
- Alex
--
UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Alex, In your description below I can see no way to - set a value of a public field of an object, or - set a value of a public field of a class Is this something that you may add later? /Jon On 04-11-10 19:53 , Alexander Burger wrote: Hi all, here is an update. Besides the basic I/O system and half-baked network functions (until now only 'connect', 'accept' and a partial 'port'), we have now a set of functions interfacing to the Java Reflection API. I'd like to describe them shortly, and hear your opinions. You can try it if you like if you download the current testing release. The central function is called 'java'. It comes in three forms: (java 'cls 'T 'any ..) -> obj This is a constructor call, returning a Java object. 'cls' is the name of a Java class, like "java.lang.StringBuilder". (java 'cls 'msg 'any ..) -> obj This calls a static method 'msg' for a class 'cls'. (java 'obj 'msg 'any ..) -> obj This calls a dynamic method 'msg' for an object 'obj' Example: : (setq Sb (java "java.lang.StringBuilder" T "abc")) -> $StringBuilder : (java Sb "append" (char: 44)) -> $StringBuilder : (java Sb "append" 123) -> $StringBuilder : (java Sb "toString") -> $String : (data @) -> "abc,123" The function 'data' in the last line converts back from Java objects to PicoLisp data. Then we have 'public', it supports two forms: (public 'obj 'any) -> obj Returns the value of a public field 'any' in object 'obj' (public 'cls 'any) -> obj Returns the value of a public field 'any' in class 'cls' Example: : (public "java.lang.System" "err") -> $PrintStream : (java @ "println" "Hello world") Hello world -> NIL Finally, we have a set of type conversion functions, to produce Java objects from Lisp data: (byte: 'num|sym) -> obj (char: 'num|sym) -> obj (int: 'num) -> obj (long: 'num) -> obj (big: 'num) -> obj We saw the usage of 'char:' in the StringBuilder example above. So the 'any' arguments to the 'java' function can either be - 'T' or 'NIL', then the type is boolean - A number, then it must fit in 32 bits and will be passed as 'int' - A normal symbol, then it will be passed as 'String' - A Java object, as returned by 'java' or one of the 'xxx:' conversion functions. - A list, then all elements must be of the same type and will be passed as 'Array' I hope this reflection interface is a good compromise between simplicity and usefulness. Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Jon, > machine. When I tried "(data @)" in your Example below, I got "data > -- Undefined". Did you change something lately? Will this 'java' Oops, yes. 'data' is now also covered by 'java': (java 'obj ['cnt]) -> any This correspends to the old 'data' function. (java 'obj 'msg 'any ..) -> obj (java 'cls 'msg 'any ..) -> obj (java 'cls 'T 'any ..) -> obj > function be documented somewhere? ;-) Yes, should be in the future. Let's first clear the smoke and see how useful it turns out ;-) > Do you now if it would be possible, or easy, to build a simple HTTP > server in ErsatzLisp? Or, maybe a better solution would be to This should also be possible. The only problem I see at the moment is the missing 'fork', so we need a non-forking server. We had some discussion about that already here in this list, iirc. Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Alex, I've now had the chance to test your latest ErsatzLisp, on a Linux machine. When I tried "(data @)" in your Example below, I got "data -- Undefined". Did you change something lately? Will this 'java' function be documented somewhere? ;-) Do you now if it would be possible, or easy, to build a simple HTTP server in ErsatzLisp? Or, maybe a better solution would be to combine ErsatzLisp with a standard HTTP server. Maybe a combination of ErsatzLisp and the Play framework (www.playframework.org) could be something? /Jon On 04-11-10 19:53 , Alexander Burger wrote: Hi all, here is an update. Besides the basic I/O system and half-baked network functions (until now only 'connect', 'accept' and a partial 'port'), we have now a set of functions interfacing to the Java Reflection API. I'd like to describe them shortly, and hear your opinions. You can try it if you like if you download the current testing release. The central function is called 'java'. It comes in three forms: (java 'cls 'T 'any ..) -> obj This is a constructor call, returning a Java object. 'cls' is the name of a Java class, like "java.lang.StringBuilder". (java 'cls 'msg 'any ..) -> obj This calls a static method 'msg' for a class 'cls'. (java 'obj 'msg 'any ..) -> obj This calls a dynamic method 'msg' for an object 'obj' Example: : (setq Sb (java "java.lang.StringBuilder" T "abc")) -> $StringBuilder : (java Sb "append" (char: 44)) -> $StringBuilder : (java Sb "append" 123) -> $StringBuilder : (java Sb "toString") -> $String : (data @) -> "abc,123" The function 'data' in the last line converts back from Java objects to PicoLisp data. Then we have 'public', it supports two forms: (public 'obj 'any) -> obj Returns the value of a public field 'any' in object 'obj' (public 'cls 'any) -> obj Returns the value of a public field 'any' in class 'cls' Example: : (public "java.lang.System" "err") -> $PrintStream : (java @ "println" "Hello world") Hello world -> NIL Finally, we have a set of type conversion functions, to produce Java objects from Lisp data: (byte: 'num|sym) -> obj (char: 'num|sym) -> obj (int: 'num) -> obj (long: 'num) -> obj (big: 'num) -> obj We saw the usage of 'char:' in the StringBuilder example above. So the 'any' arguments to the 'java' function can either be - 'T' or 'NIL', then the type is boolean - A number, then it must fit in 32 bits and will be passed as 'int' - A normal symbol, then it will be passed as 'String' - A Java object, as returned by 'java' or one of the 'xxx:' conversion functions. - A list, then all elements must be of the same type and will be passed as 'Array' I hope this reflection interface is a good compromise between simplicity and usefulness. Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Done! -- Was: Announce: ErsatzLisp (Java PicoLisp)
Hi Jon, > I look forward to try this ErsatzLisp, but I have no Java 1.6 available > yet. I tried to install it on Ubuntu, but I didn't succeed. I ended up > with a broken package, and haven't managed to clean up the mess yet. ;-) Oh - too bad! Sometimes everything goes wrong. > Maybe you could write a note on this ErsatzLisp on the Forum News > section here: http://picolisp.com/5000/-2-B.html Yes, good idea. That's what that page is for after all. > I'm also curious about "bootstrapping the 64-bit version". What's in > it? I wrote a little about that in http://software-lab.de/INSTALL and http://software-lab.de/doc64/README. In a nutshell: While Ersatz PicoLisp should work right of the box (perhaps an illusion as we see in your case ;-), and Pil32 just needs a C compiler to build, does Pil64 need a running PicoLisp system to build the "*.s" files from the sources "src64/*.l". Now the build process started by (cd src64; make picolisp) is clever enough to fall back to "ersatz/picolisp" if there is no "bin/picolisp" yet. That's all. This assumes, however, that a Java runtime system is installed. If not, the pre-generated "*.s" files can be downloaded. The rationale behind that is: These "*.s" files are relatively large, and quite redundant as they can be easily built from the "src64/*.l" files. Until now we have only "*.s" files for the 'x86-64' architecture and 'linux' platform. But if there are more in the future, the release size would explode. Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Done! -- Was: Announce: ErsatzLisp (Java PicoLisp)
On Fri, 12 Nov 2010 20:04:01 +0100, Alexander Burger
wrote:
> On Sat, Oct 30, 2010 at 07:07:01PM +0200, Alexander Burger wrote:
>> In the near future I will continue, trying to supply as many missing
>> parts as possible. The next will probably be the *Run/task
>> mechanisms
>> (using java.nio.channels.Selector), remaining I/O functions, and
>> networking.
>
> Done! I do now consider Ersatz PicoLisp complete :)
>
> At least one of the primary purposes, bootstrapping the 64-bit
> version,
> is achieved. As a consequence I removed the rather large
> pre-generated
> "*.s" files from the distribution. They will now be automatically
> build
> by 'make' if a Java runtime system is installed. If not, they are
> available for download ("http://software-lab.de/x86-64.linux.tgz";).
>
> Most things work surprisingly well, and better than I initially
> expected. "ersatz/README" tells a little more.
>
> Cheers,
> - Alex
Hi Alex,
I look forward to try this ErsatzLisp, but I have no Java 1.6 available
yet. I tried to install it on Ubuntu, but I didn't succeed. I ended up
with a broken package, and haven't managed to clean up the mess yet. ;-)
Maybe you could write a note on this ErsatzLisp on the Forum News
section here: http://picolisp.com/5000/-2-B.html
I'm also curious about "bootstrapping the 64-bit version". What's in
it?
/Jon
--
UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Done! -- Was: Announce: ErsatzLisp (Java PicoLisp)
Hi Jakob, > Totally cool! I consider the main benefit that we have a PicoLisp in > Java, which we need to be hype compliant. :) Very true ;-) -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Done! -- Was: Announce: ErsatzLisp (Java PicoLisp)
On 11/12/2010 08:04 PM, Alexander Burger wrote: [...] > Done! I do now consider Ersatz PicoLisp complete :) Totally cool! I consider the main benefit that we have a PicoLisp in Java, which we need to be hype compliant. :) // Jakob -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Done! -- Was: Announce: ErsatzLisp (Java PicoLisp)
On Sat, Oct 30, 2010 at 07:07:01PM +0200, Alexander Burger wrote:
> In the near future I will continue, trying to supply as many missing
> parts as possible. The next will probably be the *Run/task mechanisms
> (using java.nio.channels.Selector), remaining I/O functions, and
> networking.
Done! I do now consider Ersatz PicoLisp complete :)
At least one of the primary purposes, bootstrapping the 64-bit version,
is achieved. As a consequence I removed the rather large pre-generated
"*.s" files from the distribution. They will now be automatically build
by 'make' if a Java runtime system is installed. If not, they are
available for download ("http://software-lab.de/x86-64.linux.tgz";).
Most things work surprisingly well, and better than I initially
expected. "ersatz/README" tells a little more.
Cheers,
- Alex
--
UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Jon, > In PicoLisp, it is legal to compare data items of arbitrary type. > How do ersatz compare arbitrary Java classes and objects to other > PicoLisp types? Are they less than T? Less than Numbers? They are larger than numbers, and less than T, because they are normal anonymous symbols (like you create with 'box' or 'new'). They just happen to carry a Java object as an attachment, and when printed they don't show something as $12345 like other anonymous symbols, but $Class. Comparing them to each other follows the same mechanism as other anonymous symbols, i.e. they use the hashCode(). The C and asm versions of PicoLisp use the address for that. An interesting feature is perhaps that you can give these symbols another name (with 'name', the same way you can change the name of anonymous and transient symbols in PicoLisp). Example: : (setq S (java "java.lang.String" T "Hello world")) -> $String : (data S) -> "Hello world" : (name S "Hello World Symbol") -> "Hello World Symbol" : S -> "Hello World Symbol" : (data S) -> "Hello world" So the name is not relevant for the Java object attachment. If the symbol has no name (is anonymous), then the class of the attachment is printed. Otherwise, the normal name is printed, but the symbol doesn't lose its "nature" as a Java object. Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
On 11/5/10 10:56 AM, Alexander Burger wrote: Hi Jon, This looks very interesting. However, I have some problems building ersatz on my Intel Mac: ..:picoLisp jkleiser$ (cd ersatz; ./mkJar) PicoLisp.java:15: cannot find symbol symbol : class Console I see. java.io.Console is avaliable only in Java 1.6 or higher. That's why I wrote that in "ersatz/README". Too bad. Is there no up-to-date JRE available for the Mac? Cheers, - Alex Hi Alex, I may have to upgrade the OS on this Mac to 10.6.something. Another ersatz question: In PicoLisp, it is legal to compare data items of arbitrary type. How do ersatz compare arbitrary Java classes and objects to other PicoLisp types? Are they less than T? Less than Numbers? /Jon -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
On 11/05/2010 10:56 AM, Alexander Burger wrote: > > Too bad. Is there no up-to-date JRE available for the Mac? Should be: http://discussions.apple.com/thread.jspa?threadID=2315116&tstart=1 http://www.piranhamethod.com/2010/04/installing-jre-1-6-on-your-mac-os-x-leopard-10-5-8-or-later/ // Jakob -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Jon, > This looks very interesting. However, I have some problems building > ersatz on my Intel Mac: > > ..:picoLisp jkleiser$ (cd ersatz; ./mkJar) > PicoLisp.java:15: cannot find symbol > symbol : class Console I see. java.io.Console is avaliable only in Java 1.6 or higher. That's why I wrote that in "ersatz/README". Too bad. Is there no up-to-date JRE available for the Mac? Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Alex, This looks very interesting. However, I have some problems building ersatz on my Intel Mac: ..:picoLisp jkleiser$ (cd ersatz; ./mkJar) PicoLisp.java:15: cannot find symbol symbol : class Console location: class PicoLisp final static Console Term = System.console(); ^ PicoLisp.java:15: cannot find symbol symbol : method console() location: class java.lang.System final static Console Term = System.console(); ^ 2 errors /Jon On 11/4/10 7:53 PM, Alexander Burger wrote: Hi all, here is an update. Besides the basic I/O system and half-baked network functions (until now only 'connect', 'accept' and a partial 'port'), we have now a set of functions interfacing to the Java Reflection API. I'd like to describe them shortly, and hear your opinions. You can try it if you like if you download the current testing release. The central function is called 'java'. It comes in three forms: (java 'cls 'T 'any ..) -> obj This is a constructor call, returning a Java object. 'cls' is the name of a Java class, like "java.lang.StringBuilder". (java 'cls 'msg 'any ..) -> obj This calls a static method 'msg' for a class 'cls'. (java 'obj 'msg 'any ..) -> obj This calls a dynamic method 'msg' for an object 'obj' Example: : (setq Sb (java "java.lang.StringBuilder" T "abc")) -> $StringBuilder : (java Sb "append" (char: 44)) -> $StringBuilder : (java Sb "append" 123) -> $StringBuilder : (java Sb "toString") -> $String : (data @) -> "abc,123" The function 'data' in the last line converts back from Java objects to PicoLisp data. Then we have 'public', it supports two forms: (public 'obj 'any) -> obj Returns the value of a public field 'any' in object 'obj' (public 'cls 'any) -> obj Returns the value of a public field 'any' in class 'cls' Example: : (public "java.lang.System" "err") -> $PrintStream : (java @ "println" "Hello world") Hello world -> NIL Finally, we have a set of type conversion functions, to produce Java objects from Lisp data: (byte: 'num|sym) -> obj (char: 'num|sym) -> obj (int: 'num) -> obj (long: 'num) -> obj (big: 'num) -> obj We saw the usage of 'char:' in the StringBuilder example above. So the 'any' arguments to the 'java' function can either be - 'T' or 'NIL', then the type is boolean - A number, then it must fit in 32 bits and will be passed as 'int' - A normal symbol, then it will be passed as 'String' - A Java object, as returned by 'java' or one of the 'xxx:' conversion functions. - A list, then all elements must be of the same type and will be passed as 'Array' I hope this reflection interface is a good compromise between simplicity and usefulness. Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi all, here is an update. Besides the basic I/O system and half-baked network functions (until now only 'connect', 'accept' and a partial 'port'), we have now a set of functions interfacing to the Java Reflection API. I'd like to describe them shortly, and hear your opinions. You can try it if you like if you download the current testing release. The central function is called 'java'. It comes in three forms: (java 'cls 'T 'any ..) -> obj This is a constructor call, returning a Java object. 'cls' is the name of a Java class, like "java.lang.StringBuilder". (java 'cls 'msg 'any ..) -> obj This calls a static method 'msg' for a class 'cls'. (java 'obj 'msg 'any ..) -> obj This calls a dynamic method 'msg' for an object 'obj' Example: : (setq Sb (java "java.lang.StringBuilder" T "abc")) -> $StringBuilder : (java Sb "append" (char: 44)) -> $StringBuilder : (java Sb "append" 123) -> $StringBuilder : (java Sb "toString") -> $String : (data @) -> "abc,123" The function 'data' in the last line converts back from Java objects to PicoLisp data. Then we have 'public', it supports two forms: (public 'obj 'any) -> obj Returns the value of a public field 'any' in object 'obj' (public 'cls 'any) -> obj Returns the value of a public field 'any' in class 'cls' Example: : (public "java.lang.System" "err") -> $PrintStream : (java @ "println" "Hello world") Hello world -> NIL Finally, we have a set of type conversion functions, to produce Java objects from Lisp data: (byte: 'num|sym) -> obj (char: 'num|sym) -> obj (int: 'num) -> obj (long: 'num) -> obj (big: 'num) -> obj We saw the usage of 'char:' in the StringBuilder example above. So the 'any' arguments to the 'java' function can either be - 'T' or 'NIL', then the type is boolean - A number, then it must fit in 32 bits and will be passed as 'int' - A normal symbol, then it will be passed as 'String' - A Java object, as returned by 'java' or one of the 'xxx:' conversion functions. - A list, then all elements must be of the same type and will be passed as 'Array' I hope this reflection interface is a good compromise between simplicity and usefulness. Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
Hi Jakob, > I don't think you need to include fork. Yeah, it is a bit heavy. It would mean to start a new thread, and copy all Lisp data. Not really difficult though. > Database support should be > possible, I guess, but I am not sure about locking. Locking _might_ work with "java.nio.channels.FileLock", but I remember bad experiences with that under Cygwin. Cheers, - Alex -- UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Re: Announce: ErsatzLisp (Java PicoLisp)
On Sat, Oct 30, 2010 at 07:07:01PM +0200, Alexander Burger wrote:
> Hi all,
>
> the Java version of PicoLisp, which is part of the testing release since
> a few weeks, is now in a state where I dare to say that the language is
> usable (just the plain language).
Very cool!
> In the near future I will continue, trying to supply as many missing
> parts as possible. The next will probably be the *Run/task mechanisms
> (using java.nio.channels.Selector), remaining I/O functions, and
> networking. What seems definitely impossible in plain Java is raw
> console input ('key', line editing). About other parts like 'fork' and
> database support I'm not sure yet.
I don't think you need to include fork. Database support should be
possible, I guess, but I am not sure about locking.
// Jakob
--
UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
Announce: ErsatzLisp (Java PicoLisp)
Hi all,
the Java version of PicoLisp, which is part of the testing release since
a few weeks, is now in a state where I dare to say that the language is
usable (just the plain language).
In the near future I will continue, trying to supply as many missing
parts as possible. The next will probably be the *Run/task mechanisms
(using java.nio.channels.Selector), remaining I/O functions, and
networking. What seems definitely impossible in plain Java is raw
console input ('key', line editing). About other parts like 'fork' and
database support I'm not sure yet.
It resides in the "ersatz/" directory. More infos in "ersatz/README".
Cheers,
- Alex
--
UNSUBSCRIBE: mailto:picol...@software-lab.de?subject=unsubscribe
