zorba-coders team mailing list archive

[Nicolae Brinza <nbrinza@xxxxxxxxx>]

Markos, I've fixed the following issues you have reported (with changes merged into this branch):

-- Also, the grammar rule for LiteralFunctionItem should use EQNAME, instead of QNAME.

I've looked into it and the QNAME production actually covers EQNAMEs. I haven't tested it but they should work, as all the other grammar rules use the QNAME production.

-- In typeops.cpp:749, you should not be casting AnyFunctionXQType to FunctionXQType.

Fixed.

-- is_equal() has not been implemented for function types. If you think it will never be called, then at least put an assertion to make sure that it will indeed never be called.

I have implemented it.

-- I think we should completely remove the option to enable/disable the HOF feature.

I have removed it from the features system.


-- There are also several failures in fn-function-lookup and misc-HigherOrderFunctions test sets.
Are you aware of these failures?

Yes, I'm aware. As I have said in an e-mail, the remaining failures are caused by the function-lookup() which I'm working on. On the other hand, there should be no failures in misc-HigherOrderFunctions, unless there are functions which use the $$context-position or $$context-size. I'll look into this.

I will update the test/fots/CMakeLists.txt but it will take a while as FOTS has no progression registration mechanism as the rbkt testdriver has. 


-- In translator.cpp:3356 what is the reason for the arity checks for fn:number and fn:static-base-uri functions?

I've looked into it and also run some tests and seems the checks are indeed redundant. I had copied the arity verification from the FunctionCall end_visit() translation, see e.g. translator.cpp:10875 in the Trunk branch. 

I have removed those checks.

-- 
https://code.launchpad.net/~zorba-coders/zorba/hof-merge/+merge/151673
Your team Zorba Coders is subscribed to branch lp:zorba.