asterixdb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Taewoo Kim <wangs...@gmail.com>
Subject Re: ATypeTag.UUID_STRING vs ATypeTag.UUID
Date Fri, 22 Jan 2016 17:40:09 GMT
@Till: I agree with you. Actually, at that time, AUUID
serializer/deserializer was already there so I thought we might need to
handle both cases (two long values / canonical representation) for
input/output. I couldn't find a way to use two serializer/deserializers
(canonical representation / two long values) in one type. Still, I don't
see how it can be done. But, it would be really nice if we can integrate
them into one.

Best,
Taewoo

On Fri, Jan 22, 2016 at 8:49 AM, Till Westmann <tillw@apache.org> wrote:

> Hi Taewoo,
>
> I’m still a little confused. To me the type tag is about the type of the
> value and it is independent of the serialization. I think that the
> serialization should be orthogonal to the type of the value. I.e. we can
> have
> a) the binary serialization/deserialization that we use internally to
> ship/store UUIDs in AsterixDB or we can have
> b) the string serialization/deserialization that is used when parsing
> external data or printing results.
> So in the case of UUIDs we have AUUIDSerializerDeserialize for a) and code
> for b) in AUUID.fromString, AUUIDFromStringConstructorDescriptor,
> AMutableUUID.fromStringToAMuatbleUUID, and AUUIDPrinter.
> So I think that we should be covered (actually a little but too much ..)
> for both cases and it seems to me that we should be able to remove
> ATypeTag.UUID_STRING without causing damage.
>
> Would you agree or is there something I’m missing?
>
> Cheers,
> Till
>
>
> On 21 Jan 2016, at 11:42, Taewoo Kim wrote:
>
> One more clarification: I have created UUID_STRING for mainly Serializer -
>> deserializer issues. Since UUID serializer/deserializer assumes that two
>> long values. It was not sufficient to serialize/deserialize a canonical
>> string notation. So, when parsing the outside data, UUID_STRING is used to
>> serialize/deserialize a UUID. And internally, we use UUID, not
>> UUID_STRING.
>>
>>
>> Best,
>> Taewoo
>>
>> On Thu, Jan 21, 2016 at 11:36 AM, Taewoo Kim <wangsaeu@gmail.com> wrote:
>>
>> I have created UUID_string.
>>>
>>> Internally, a UUID is represented as two long values. However, generally
>>> we use a canonical string notation. So, when we parse a UUID, I had to
>>> create a type that can represent a canonical form of UUID. So,
>>> UUID_STRING
>>> mainly exists for the parsers.
>>>
>>> Best,
>>> Taewoo
>>>
>>> On Thu, Jan 21, 2016 at 11:32 AM, Till Westmann <tillw@apache.org>
>>> wrote:
>>>
>>> Hi,
>>>>
>>>> who knows
>>>> a) what the difference between ATypeTag.UUID_STRING and ATypeTag.UUID is
>>>> and
>>>> b) why we need both?
>>>>
>>>> Cheers,
>>>> Till
>>>>
>>>>
>>>
>>>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message