[Yanel-dev] [Yarep] Bug/feature re indexing of properties

[Guillaume Déflache]

Michael Wechner schrieb:
> Guillaume Déflache wrote:
>>
>> It looks like the property values also get escaped, I get:
> 
> yes (on purpose), but it should also be unescaped when reading it (in 
> the case of version 1.0), otherwise this would be a bug
> [...]
> as mentioned above these values should be unescaped. AFAIK, within the 
> tests this is working.
>>
>>
>>> Also I have fixed another problem re line breaks which was noted as 
>>> TODO within the code.
>>>
>>> Also to make it backwards compatible I have introduced a version 
>>> number which is a property itself:
>>>
>>> yarep_vfs-meta-file-version<string>:1.0
>>
>> It looks like we may need to define 1.0.1 to fix this.
> 
> fix what exactly?

I thought also escaping the value was an oversight, but now I know it isn't.


>>> which means all meta data which does not contain this property is 
>>> being handled as before, but as soon as it is being
>>> rewritten it should be handled with escaped colons.
>>
>> So AFAICU there is no workaround: as soon as one writes to the file, 
>> the new 1.0 gets used, right?
> 
> yes. Any reason why this shouldn't happen?

None I can think of ATM.


>>> All my tests are fine, but in case you should encounter any problem, 
>>> please let me know.
>>
>> All the related tests are in 
>> /home/guillaume/yarep.VCWC/src/test/java/org/wyona/yarep/tests/VirtualFilesystemRevisionsTest.java 
>> in the 'testEscapeUnescapePropertyName' method, right?
> 
> no, there is also
> 
> src/test/java/org/wyona/yarep/tests/VirtualFilesystemIndexSearchTest.java
> 
> 
>> It seems the case with ':' in property values is explicitely tested, 
>> but AFAICS only with values of type string, not with values of type 
>> date, maybe that's the problem.
> 
> I have tested this with the from scratch realm also and it seemed fine 
> to me there.
> Pleaes also test it with the from scrartch realm and let me know if this 
> should be a problem.
> 
> Maybe the root of problem is somewhere else, but you need to give me 
> more information about where exactly you are experiencing these issues.

OK sorry, it looks like after a `build clean-all` and removing partially 
written test data everything is working again as expected.

Cheers,
    Guillaume