ibatis-user-java mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From huym...@swiftdsl.com.au
Subject Re: Flexible ResultMaps (Again)
Date Mon, 07 Feb 2005 05:37:56 GMT
> Okay, I just reviewed my previous answers and perhaps we need to
> answer more clearly your questions exactly:
>
> We favour explicit behaviour over implicit.  We do so in line with
> strongly typed languages that we support such as Java and C#.  We
> detest implicit behaviour because it is difficult to maintain and
> debug.

There was another recent thread about parameter maps which asked for the
same thing which you thought was a good idea. I'm not sure what the
difference is there (in terms of implicit behaviour). (Not sure how to
provide link to this other thread)

> Trust me, your suggestion will not make things easier, it will
> only make tradeoffs that are equal at best and probably less
> desirable.

I'm sorry to sound distrustful, but if you can give me an example of where
it would not be desirable I would accept it. Sorry for soundind like a
nagging free wheeling programmer but I reallly believe sqlmap will benefit
from this from simplifying the resultmaps.

> We are considering other options for ease of maintenance
> such as supporting the <generate> tag (like the C# version) and
> possibly considering inline result maps (which are explicit, but also
> easy to maintain).

I really don't like inline result maps because you'd have alot of
duplicate    inline mappings all over your sqlmap which not only looks
messy but would result in quite a maintenance problem IMO. I hope you
never get rid of resultmaps (even without what I'm asking for).

> We're also looking into a number of tooling
> options to help with the maintenance of SQL maps.

I guess this would be a way around complex sqlmaps but I love getting my
hands dirty and would rather have the maps easy to maintain by hand.

> This is my initial answer.  My suggestion to you is to add a Wish or
> New Feature entry into JIRA so that we don't lose track of this.
> You're the second person to request this.  With some more support, we
> will prioritize it accordingly and you may yet see it in a future
> release.

Thanks for the response, I'll do this.

> Cheers,
> Clinton
>
>
>
>
>
> On Sun, 6 Feb 2005 22:02:15 -0700, Clinton Begin
> <clinton.begin@gmail.com> wrote:
>> I believe we already answered your question...this is the second
>> thread you've posted on the subject, no?
>>
>> Clinton
>>
>>
>> On Mon, 07 Feb 2005 09:13:31 +1100, Huy <huymail@swiftdsl.com.au> wrote:
>> > Clinton Begin wrote:
>> > >>>Looosly typed!
>> > >
>> > >
>> > > That's not loose typing.  It's NO typing.
>> > >
>> > > The type that's missing isn't what you think it is.  The type that's
>> > > missing is your domain class definition.  You don't have types at
>> all.
>> > >  You just have collections of implicitly related information.
>> > >
>> > > In any case, I appreciate your heavy, thorough testing of our Map
>> > > support (seriously actually).  ;-)
>> > >
>> > > Cheers,
>> > > Clinton
>> > >
>> >
>> > Now that you guys have concluded that map usage is not desirable ( I
>> > totally agree), is there any chance of answering my original question
>> > about the flexible resultmaps ?
>> >
>> > Thanks.
>> >
>> > Huy
>> >
>>
>


Mime
View raw message