Latest release

[UaleWonka]

Hi,

I am using Windows 10, and I launched ConlluEditor with the jar in the .zip of the latest release.

I have found 2 probable bugs:

1. I cannot launch the validator: it gives me this error:

Validator error: Cannot run program "C:\UD\tools\validate.py": CreateProcess error=193, %1 non è un'applicazione di Win32 valida

2. When I try a compose or a split operation, it doesn't modify the sentence text and it doesn't work anymore.

|incoherent "# text" and forms: «Tormenta undique conquiri et frumentum mitti auxilia adduci iubet.» ≠ «Tormenta undique conquiri etfrumentum mitti auxilia adduci iubet.|

When I use the sentsplit operation, it gives me this error too (and it cannot undo the operation):

General error: Cannot invoke "String.charAt(int)" because the return value of "com.orange.labs.conllparser.ConllSentence.getText()" is null

Thank you!

[jheinecke]

Hi, For the first error, it looks like your windows has no python installed. Since C:\UD\tools\validate.py is a python programme, you should install python. If it _is_installed, maybe changing it to python C:\UD\tools\validate.py (in the validator script) works? I'll have a look in the compose/split problem. Can you post the sentence and the operation which caused this bug?

[jheinecke]

In fact the sentsplit operation (and the compose operation) do not modify the # text ... line of the sentence. Once the sentence is split, you should use the edit metadata button to adapt the text.

[Stormur]

In fact the sentsplit operation (and the compose operation) do not modify the # text ... line of the sentence. Once the sentence is split, you should use the edit metadata button to adapt the text.

Is this a new behaviour? Or specific to the precompiled jar? I do not remember having issues with this :thinking:

[jheinecke]

It is :-( since the # text is the original text, and modifiying it automatically without asking the user could cause errors, I stopped this. I could add it as an option though.

[Stormur]

Ah! Yes, I think that many would opt for this, in my experience it has always been a sensible thing.

At least when doing sentence splitting, I think it is no use of keeping the split tokens together with their old sentence (this is part of the bug reported by the original poster).

[jheinecke]

OK, I agree, I will update the # text = ... line after sentsplit, sentjoin and compose operations, as users actively change the tokenisation, and the original sentence is of no importance any more

[UaleWonka]

Hi, For the first error, it looks like your windows has no python installed. Since C:\UD\tools\validate.py is a python programme, you should install python. If it _is_installed, maybe changing it to python C:\UD\tools\validate.py (in the validator script) works? I'll have a look in the compose/split problem. Can you post the sentence and the operation which caused this bug?

I modified the validator script as you suggested and now it works. Thank you!

About the second issue, here are the two cases:

Sentence: Bello Alexandrino conflato Caesar Rhodo atque ex Syria Ciliciaque omnem classem arcessit. Operation: split 9 Bug: incoherent "# text" and forms: «Bello Alexandrino conflato Caesar Rhodo atque ex Syria Ciliciaque omnem classem arcessit.» ≠ «Bello Alexandrino conflato Caesar Rhodo atque ex Syria Ciliciaque Ciliciaque omnem classem arcessit.|

Here I can still move to the next sentence, but I cannot undo the operation.

Sentence: (Quarum alterius rei copiam exiguam, alterius nullam omnino facultatem habebat.) Quod utrumque large palus praebere poterat. Operation: sentsplit 15 Bug: incoherent "# text" and forms: «(Quarum alterius rei copiam exiguam, alterius nullam omnino facultatem habebat.) Quod utrumque large palus praebere poterat.» ≠ «(Quarum alterius rei copiam exiguam, alterius nullam omnino facultatem habebat.)|

Here I cannot undo the operation ( it says me this: General error: Cannot invoke "String.charAt(int)" because the return value of "com.orange.labs.conllparser.ConllSentence.getText()" is null), nor I can move to next sentence (it says me this: General error: Cannot invoke "String.charAt(int)" because the return value of "com.orange.labs.conllparser.ConllSentence.getText()" is null)

Thank you again!

[UaleWonka]

I just noticed that I cannot undo anything. Even if I modify a dependency relation, when I try to undo it, it says me:

General error: Cannot invoke "String.charAt(int)" because the return value of "com.orange.labs.conllparser.ConllSentence.getText()" is null

[jheinecke]

Hi, I just released version 2.22.0, which corrects this bug (and also gives the possibility to do shortcuts for values in the MISC column (@Stormur :-) ) you can either do git pull and mvn install or grab the V2.22.0 release

[UaleWonka]

Hi, I just released version 2.22.0, which corrects this bug (and also gives the possibility to do shortcuts for values in the MISC column (@Stormur :-) ) you can either do git pull and mvn install or grab the V2.22.0 release

Thank you! Now the sentsplit and the split operations work.

I still have problems with the undo button. Whatever I do (modifying deprels, or using operations), I cannot undo them. It always says me:

General error: Cannot invoke "String.length()" because the return value of "com.orange.labs.conllparser.ConllSentence.getText()" is null

The same error message pops out when I move to another sentence, and I try to read again the sentence I modified before.

[jheinecke]

I found an error in my unitary tests, so this bug got through. It should be OK now with V2.22.1

[UaleWonka]

I found an error in my unitary tests, so this bug got through. It should be OK now with V2.22.1

Now it works perfectly. Thank you for your support!

[UaleWonka]

Hi, I just released version 2.22.0, which corrects this bug (and also gives the possibility to do shortcuts for values in the MISC column (@Stormur :-) ) you can either do git pull and mvn install or grab the V2.22.0 release

Sorry for bothering you again.

I'm using the shortcuts in the MISC column, but when I use it on a node which already contains the annotation SpaceAfter=No, it deletes it and replace it with my annotation, thus leading to an error (incoherent "# text" and forms).

Could you please help me?

[jheinecke]

Thanks for pointing me to this, it is corrected in V2.22.2 (git pull && mvn install or latest release)

[UaleWonka]

Thanks for pointing me to this, it is corrected in V2.22.2 (git pull && mvn install or latest release)

Thank you!