[Seaside] [RFT] New web project

Jason Johnson jbjohns at libsource.com
Fri Nov 10 21:17:26 UTC 2006


William Harford wrote:
>
> On Nov 9, 2006, at 3:11 PM, Jason Johnson wrote:
>
> I mean you build/change/extend the application in the application via 
> dragging, clicking, grabbing halos, etc.
Yes, we are in line on this point.

> Yes I mean hierarchical. There are ways to store hierarchical data in 
> relational database. I created a project called REServe that does just 
> that ( http://squeaksource.com/REServe.html ) but you sacrifice some 
> of the advantages of both models by doing so. It's a trade off.
>
> It makes life easier for users because it's one less thing to think 
> about. If you are developing a Smalltalk application it is inherently 
> hierarchical. If you want to persist objects in a relational database 
> some tradeoffs will have to be made and you loose some of the power 
> that Smalltalk gives you.  For example polymorphism ( REServe works 
> around this problem by having a lookup table and unique IDs for every 
> row in the database regardless of table) .
Hrm.  This I am less sold on at the moment.  I think the main use cases 
for a database are going to be data that isn't modeled in smalltalk.  I 
guess the classic example would be HR information about employees.  You 
have a table for the person with their address, date of birth, position, 
department, etc.  You have a table of possible positions, possible 
departments, but the point is, this data looks very relational to me.  I 
wouldn't want to lose any of the power that relational gives me unless I 
got something really good in return.

Maybe you could provide a real world type case where this makes life a 
lot easier for the user?  And remember my criteria for a good web 
framework is the further you can go without writing code the better the 
frame work.

> The idea is that you limit if not completely eliminate the abstraction 
> between developing the application and using it. MS Access sort of 
> does this although I don't think it's the best example.
>
> I say think way outside the box on this.
> You stand a better chance of the project getting attention by doing 
> something radical :-)
Oh, I intend for us to stick out. :)  But I'm a little confused on this 
point still.  I kind of saw the pinnacle of GUI programming as MVP, 
which is an abstraction.  Are you saying the way to go is to abandon the 
abstractions and put that kind of knowledge back with the data, etc.?  
As far as I know MS Access, VB, Borland and all those guys try to do a 
kind of MVC but they jam the M and the VC so closely together that your 
data ends up hopelessly bound to display.  Which might be good if you 
write a little tiny app, but the more complicated your data gets the 
more abstractions help.  But maybe what you're saying is so radical I 
havn't gotten it yet. :)

And I hope I am not coming off as being critical of you or anyone else's 
ideas.  I am glad to see ideas expressed so we can, hopefully, end up 
with a really advanced system.  But I just have to keep an eye out for 
the common case and the common user.  Nothing is worse then a big fancy 
feature that no one needed.  So I will have to hold up any ideas any of 
us come up with to this standard.  I haven't decided any ideas I have 
seen so far fit in this group, but there are a few things I want to 
understand better.



More information about the Seaside mailing list