Re: [gobo-eiffel-develop] Catcall in gec

[Eric B. <er...@go...>]

Wolfgang Jansen wrote:
> Eric Bezault wrote:
>> Wolfgang Jansen wrote:
>>> Hi,
>>>
>>> when experimenting with classes of the GEC it happens
>>> that compilation fails because of a catcall.
>>> To be precise, let `f' be a ET_DYNAMIC_FEATURE
>>> then adding the line
>>>
>>>    if f.static_feature.name.name.is_equal("abc") then ... end
>>>
>>> leads to compiler message
>>>
>>> [CATCALL] class ET_C_GENERATOR (17182,4): type 'STRING_8' of actual
>>> argument #1 does not conform to type 'UC_UTF8_STRING' of formal 
>>> argument in feature `is_equal' in class 'UC_UTF8_STRING'
>> Yes, the signature of `is_equal' is a real problem when
>> we want to be CAT-call free. The workaround is to inherit
>> from KL_IMPORTED_STRING_ROUTINES and write:
>>
>>   if STRING_.same_string (f.static_feature.name.name, "abc") then
>>
>> Another solution:
>>
>>   if f.static_feature.name.same_feature_name (create 
>> {ET_IDENTIFIER}.make ("abc")) then
>>
>>
>>> A similar catcall occurs if a DS_HASH_TABLE[INTEGER,STRING]
>>> object is created (a call at the object is not needed).
>> For hash-tables, you need to create them that way in order
>> to be CAT-call free:
>>
>>   create ht.make_map (...)
>>   ht.set_key_equality_tester (string_equality_tester)
>>
>> or:
>>
>>   create ht.make_with_equality_testers (..., Void, 
>> string_equality_tester)
>>
>> where `string_equality_tester' is inherited from
>> KL_SHARED_STRING_EQUALITY_TESTER.
>>
>>
> Thanks, it works (though it is hard to understand).

Running gelint with the '--cat' option will tell you the
code chain that could lead to the CAT-call. This might look
too constraining to have theses compilation errors, but it's
probably better to have them at compilation time than at
run-time. At run-time, it's even harder to understand.

BTW, I assumed you knew what a CAT-call is. If you don't,
let me know and I'll show you a small example.

-- 
Eric Bezault
mailto:er...@go...
http://www.gobosoft.com