Re: Yet another idea

["Alexey Parshin" <alexeyp@gmail.com>]

Most likely, I'd have to create a view for UMO with the needed fields.
 Topic is a UMO. So, the type would by CUMO, for instance.

2008/7/1 Ilya A. Volynets-Evenbakh <ilya@total-knowledge.com>:
> Let me repeat the question: what is the data type of "topics" variable?
>
> Alexey Parshin wrote:
>>
>> If the list isn't result of the query - it's not related to the
>> database and should be static.
>> Most likely, I'd have wrappers for all the queries - to keep 'em
>> auto-prepared and re-used.
>>
>> Topic is a UMO. UMOs are stored in umo_version table. There is no name
>> or title for a UMO in that table, so we need some query to get it.
>> Otherwise, it's trivial.
>>
>> 2008/7/1 Ilya A. Volynets-Evenbakh <ilya@total-knowledge.com>:
>>
>>>
>>> Where is the guarantee that every object list is direct result of a
>>> query?
>>> Plus, will you have a separate wrapper for every query?
>>>
>>> IOW: what class is "topics" object in your example?
>>>
>>> Alexey Parshin wrote:
>>>
>>>>
>>>> I'm seeing a bit differently. The implementation should use not the
>>>> members of the class but fields in the underlying query. And the
>>>> parser should just once per loop, converting a template into a vector
>>>> of pairs of text an following field index, so the template:
>>>>
>>>> <A href='HomeServlet?id={uv_id}'> {uv_name} </A>
>>>>
>>>> is parsed (once, before the loop) into vector of pairs:
>>>>
>>>> { "<A href='HomeServlet?id=", 0 },
>>>> { "'> ", 1 },
>>>> { "</A>", -1}
>>>>
>>>> The following loop would work really fast cause it doesn't parse
>>>> anything.
>>>>
>>>> Alexey
>>>>
>>>> 1 июля 2008 г. 16:12 пользователь Ilya A. Volynets-Evenbakh
>>>> <ilya@total-knowledge.com> написал:
>>>>
>>>>
>>>>>
>>>>> In other words, what you are suggesting, is writing wrapper class for
>>>>> each
>>>>> possible UMO collection, that would allow easy printing of enumerations
>>>>> of such objects?
>>>>>
>>>>> That may be a good idea, although parsing such templates will be
>>>>> time consuming, I'm afraid. Another obvious problem is maintaining
>>>>> such wrappers, whenever object structure changes (e.g. a new field
>>>>> is added to UMO and needs to be exposed to list generators).
>>>>>
>>>>>  Ilya.
>>>>>
>>>>> Alexey Parshin wrote:
>>>>>
>>>>>
>>>>>>
>>>>>> Илья,
>>>>>>
>>>>>> Я тут подумал над тем, как бы избавиться от C++ циклов в .csp, в
>>>>>> пользу
>>>>>> чего-нибудь более прозрачного, и вот что придумал. Почему бы нам,
>>>>>> вместо
>>>>>> циклов, не предоставить генерацию по шаблону.
>>>>>>
>>>>>> Вместо:
>>>>>>
>>>>>> for (vector<CTopic>::iterator itor = topics.begin(); itor !=
>>>>>> topic.end();
>>>>>> itor++)
>>>>>>  out << "<A href='TopicServlet?topic_id=" << (*itor)->id << ">"  <<
>>>>>> itor->title << "'</A>";
>>>>>>
>>>>>> сделаем более простое и прозрачное:
>>>>>>
>>>>>> topics.forEach("<A href='TopicServlet?topic_id={id}> {title} </A>");
>>>>>>
>>>>>> --
>>>>>> Alexey Parshin,
>>>>>> http://www.sptk.net
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>
>>
>>
>>
>
>



-- 
Alexey Parshin,
http://www.sptk.net