[Seaside-dev] codecs overhauled (again)

[Julian Fitzell]

On Tue, Jun 9, 2009 at 2:27 AM, Philippe
Marschall<philippe.marschall at gmail.com> wrote:
> 2009/6/7 Julian Fitzell <jfitzell at gmail.com>:
>> Agreed. I also would still like this protocol to be expressed in terms
>> of both an internal and external encoding pair (which I think (?) we
>> managed to agree to in principle at the last sprint).
>>
>> Maybe we don't want to fully implement functionality that uses that
>> right now but it might still be a good idea to lay the groundwork for
>> it now.
>>
>> WACodec external: 'utf-8' internal: 'utf-8'.  "an instance of WANullCodec"
>> WACodec external: 'latin-1' internal: 'latin-1'.  "an instance of WANullCodec"
>> WACodec external: 'utf-16' internal: 'utf-16'.  "an instance of WANullCodec"
>>
>> WACodec external: 'utf-8' internal: nil.  "an instance of WAUtf8Codec"
>> (maybe there's a better value than nil for "native"? #native?)
>> WACodec external: 'utf-8'. "equivalent to above"
>>
>> WACodec external: 'latin-1'.  "an instance of WAGenericCodec"
>>
>> WACodec external: 'utf-16' internal: 'utf-8'.  "Error: no codec
>> supports that encoding pair"
>>
>> I know this is my agenda item I keep pushing but I do think it will
>> help users understand what input and output they should be expecting.
>> I can try to find time to do this if you don't want to but I thought
>> I'd bring it up again since you were in there anyway.
>>
>> Also, on another note, do we not want to keep tests that explicitly
>> test WAGenericCodec? I see that it's nice to have generic UTF-8 tests
>> that just test that there is *some* codec that will do the conversion
>> but doesn't it still make sense in the Squeak tests to make sure that
>> the generic codec is functioning properly? Right now the tests all use
>> "WACodec forEncoding:" which might not even return a WAGenericCodec.
>
> So? I mean they test that the system is working. Isn't it OO to not
> care or know which class and code ends up doing the work?

It may be OO but it's not necessarily Unit Testy. I agree it's useful
to have a test of "does UTF-8 decoding work" but it's also useful to
have a test of "does this class work". WAGenericCodec has some more
complex behaviour since it allows various combinations of encodings
and depends on underlying squeak stuff that we should be exercising, I
think.

Julian