On 6 August 2012 20:28, Mikel Artetxe <[email protected]> wrote:
> On Mon, Aug 6, 2012 at 4:41 PM, Jimmy O'Regan <[email protected]> wrote:
>>
>> Thanks to the intrepid bug hunting work of Mikel Artetxe and Kevin
>> Unhammer (and, indirectly, Jacob Nordfalk, via lttoolbox-java's error
>> reporting), I've made bugfix releases available for the following
>> packages:
>>
>> apertium-es-ro (0.7.3)
>> apertium-oc-ca (1.0.6)
>> apertium-oc-es (1.0.6)
>> apertium-pt-gl (0.9.2)
>>
>> Testing is required for all of them, as all involved fixes to rule
>> syntax - while fixing the rules and macros to work as the author had
>> intended *should* lead to an overall improvement, if work-arounds were
>> found, it's possible there may be degradation. So, please, try them
>> out and let us know.
>>
>> In addition, I've also made available a version of apertium-es-ast
>> which removes the dependency on apertium-pn-recogniser:
>> apertium-es-ast_NO-PN-1.1.0.tar.gz
>
>
>
> Thank you very much! I've published the JARs for all of them except
> apertium-es-ast, so we now have 24 working pairs out of the 31 released
> ones! However, I still have some problems...
>
> For the Occitan pairs (both apertium-oc-ca and apertium-oc-es),
> lttoolbox-java fails to generate bytecode for the transfer files of the
> Aranese dialect, so the released JARs don't contain the modes that
> correspond to it. I get something like this for all of them:
>
> trouble processing:
> class name (transfer_classes/es_oc_aran_t1x) does not match path
> (transfer_classes/es_oc@aran_t1x.class)
> ...while parsing transfer_classes/es_oc@aran_t1x.class
> ...while processing transfer_classes/es_oc@aran_t1x.class
>
> I guess that the problem is the "@" symbol at the filename... What should we
> do, rename the files so that they don't contain the at symbol or adapt
> lttoolbox-java so that it replaces them by something that is allowed in Java
> class names (an underscore, for instance)?
>
Replace '@' with '_' and it should be fine.
>
> As for apertium-es-ast, the bytecode is generated without any problem, but
> then I get the following error when trying to load it:
>
> Exception in thread "Thread-25" java.lang.ClassFormatError: Invalid method
> Code length 66654 in class file transfer_classes/apertium_es_ast_es_ast_t1x
>
> I suspect that the problem is related to my code that generates the bytecode
> directly through BCEL (or perhaps it's a bug in BCEL itself, I did have a
> similar problem that happened to be caused by a BCEL bug...). So, well, I
> will try to fix it...
The older version of lttoolbox-java gives:
Compiling: javac -cp
/Users/jim/Apertium/lttoolbox-java/dist/lttoolbox.jar
./apertium_es_ast_es_ast_t1x.java
./apertium_es_ast_es_ast_t1x.java:20071: code too large
public void rule98__proclise__prnpro__prnpro__haber__verb(Writer out,
TransferWord word1, String blank1, TransferWord word2, String blank2,
TransferWord word3, String blank3, TransferWord word4, String blank4,
TransferWord word5) throws IOException
^
./apertium_es_ast_es_ast_t1x.java:39390: code too large
public void
rule120__proclise__prnpro__prnpro__haber__verbir__a__inf_ger(Writer
out, TransferWord word1, String blank1, TransferWord word2, String
blank2, TransferWord word3, String blank3, TransferWord word4, String
blank4, TransferWord word5, String blank5, TransferWord word6, String
blank6, TransferWord word7) throws IOException
...if that helps at all.
--
<Sefam> Are any of the mentors around?
<jimregan> yes, they're the ones trolling you
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Apertium-stuff mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/apertium-stuff
