Dear Pieter,

This sounds as what I was looking for (but I'll keep the good tip of
Erik and Bas in mind if I have to exchange more OO oriented messages,
which I undoubtedly going to do in the near future...) Going to
experiment with generics then!

However, where can I find this slightly extended TCP interface from
Clean (and is it also available for linux?), you say it is in the
distribution, but in what file (there are quite some TCP related files
in the lib directory). Moreover, it would be great if you have an
example application showing such a test.

________________________________
Van: [email protected] [[email protected]] namens 
Pieter Koopman [[email protected]]
Verzonden: maandag 21 november 2011 16:05
Aan: [email protected]
Onderwerp: Re: [clean-list] Dynamic typing for communication with 
non-cleanprograms

Dear Chide,

In order to test programs written in Java with Gast I use a slightly extended 
TCP interface from Clean. There is a TCP interface in distribution (or rather 
two of them). The messages are arbitrary data types that are encode en decode 
on the Clean side by some generic functions. On the Java side the user has to 
do some work on encoding decoding manually. This works fine for the rather 
simple messages used by Gast..

Maybe the JSON stuff works even better since there is a better interface on the 
Java side. On the Clean side there is also a generic interface, you can just 
derive the needed conversions. I have no experience with this JSON library 
(apart from using it in iTasks).

Dynamics would not be my choice since they contain unevaluated Clean 
expressions and (references to) Clean code. You would need too much machinery 
to handle this in any other language.

Best, Pieter

On 21/11/2011 3:02 PM, Bas Lijnse wrote:
The JSON libraries that Erik mentions are available from:
https://svn.cs.ru.nl/repos/clean-platform/trunk/src/libraries/OS-Independent/Text/

I always use this when I want to do simple data exchange between Clean and 
another language (I may be biased as author/maintainer of the library though :) 
)

Best,
Bas

On 21-11-11 14:52, [email protected]<mailto:[email protected]> wrote:

And dump all the hard work on Java? Dynamics are a proprietary Clean format and 
likely more complex/liberal than you need. Maybe JSON is easier to use: 
probably available in your JAVA implementation already and under the hood JSON 
converters are already available in Clean iTasks.


-----Oorspronkelijk bericht-----
Van: [email protected]<mailto:[email protected]> 
namens Groenouwe, C.
Verzonden: ma 21-11-2011 13:39
Aan: [email protected]<mailto:[email protected]>
CC: [email protected]<mailto:[email protected]>
Onderwerp: [clean-list] Dynamic typing for communication with non-cleanprograms

Dear Pieter (and others),

Then another question:

If you want to send information from one programming language (e.g. Java) to 
Clean, a typical way to go would be the following. First define an intermediate 
representation (for example in XML, or your own format). Then write a generator 
on the sender's side (in this case in Java), which converts a Java value or 
object into a file in the intermediate representation. Finally, write a parser 
on the receiver's side (Clean) which parses file holding the intermediate 
representation into a Clean value (with a Clean type). This is a labourious 
process during which you have the feeling you are are doing a lot of 
unnecessary work. As if you are repackaging the same sandwich three times in a 
row, first in a Java structure, then in the intermediate structure, and then 
again in a Clean structure... Waste of time (and the environment...) ;-)

However, I had an idea: is it possible to use Clean's dynamic types for this 
purpose? Instead of generating some arbitrary intermediate type on the sender's 
side, immediately generate a value that can be read by Clean's dynamic type 
system. This would cut your work in half: you don't have to write a parser on 
Clean's side any more.

I don't have any experience using Clean's dynamic typing yet, therefore I 
decided to first ask whether it is a fruitful approach before exploring it. 
Thus, is this possible and is it a good idea? Of course, I'm also open to other 
suggestions.

Thanks in advance,

Chide







_______________________________________________
clean-list mailing list
[email protected]<mailto:[email protected]>
http://mailman.science.ru.nl/mailman/listinfo/clean-list





_______________________________________________
clean-list mailing list
[email protected]<mailto:[email protected]>
http://mailman.science.ru.nl/mailman/listinfo/clean-list

_______________________________________________
clean-list mailing list
[email protected]
http://mailman.science.ru.nl/mailman/listinfo/clean-list

Reply via email to