cayenne-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Richard Frovarp <rfrov...@apache.org>
Subject Re: basic tutorial
Date Sun, 23 Dec 2012 16:23:51 GMT
On 12/22/2012 05:31 PM, Michael Jaruska wrote:
> hmm, i was in hope not to make spam but instead of (short) point to 
> some notes
> about using cayenne without modeler but just hand-editing xml files 
> there was rubbish
> about xml and editing them by hand. i'm editing xml by hand for more 
> than 10 years
> without problems and there's more efective formats for machines than xml.
>
> btw: in cayenne i'm beginner, hibernate, ibatis etc i'm using for 
> years and
> yes, i (and in fact all my colleagues) am editing xml's by hand.
>
> from this point of view, programmers of modeler are programming in 
> wrong way...
>

I used to edit SQL by hand for years. I've even hand created mapping 
from SQL results into objects. I'm sure we all have. The whole point of 
ORMs is to make life easier. I believe the Cayenne community takes a 
different approach to all of this than many other systems. I haven't 
seen this explicitly state, but this is the general feeling I get.

There's a famous talk about how ORMs suck. One of my sys admins sent it 
to me a couple of years ago, but I can't find it now. In it, the 
speakers goes off on a rant about how ORMs were supposed to make life 
better, easier for the developer. Then he shows some annotation hell 
from Hibernate, and asks at what point did we decide that those 
annotations were better than straight SQL?

With Cayenne we get to skip the SQL, we get to skip writing code (as it 
is generated), and we get to skip the horrible annotations (it doesn't 
use them). That leaves us with the XML document to generate the code, to 
do our SQL. Honestly, I think I would rather do straight SQL than have 
to generate the XML. That's what you have to do in Torque. On of the 
things that makes Cayenne stand apart is that modeler. The modeler is 
accurate and quick. I can get in, do what I need to do, and get on with 
more interesting parts of my code.

In this way the developers of the modeler has done exactly what I want, 
something quick, simple, and less error prone than doing SQL, 
annotations, or XML directly.

Like others have said, if you want to do it by hand, use the modeler as 
a tutorial. You can quickly figure out what it is doing by diffing your 
changes as you go. There are times when editing by hand is advantageous, 
with the modeler providing validation. Other developers can pick up your 
changes in commit messages fairly well.

Mime
View raw message