From: Stephan Diehl <stephan_diehl@ho...> - 2001-03-11 21:00:31
playing around with webware for a few weeks now, I happy to tell, that all
in all, I like it a lot (though I still don't understand everything).
A few thoughts:
1. Since Webware has a session management, it might happen, that one is
visiting the same url twice, but in different contexts (maybe one has an
application, that collects different data and in the end one goes to an url
"myApp?action=save". Anyway, if the Browser has caching enabled, chances
are, that the browser won't even try to connect to the server since he
already "knows" the url.
I was thinking, that we needed the possibility to add a "random" attribute
to every request. This could serve two purposes: a) every url were unique
and b) one could decide very easily, if a request was valid in my web
2. The handling of session and request variables get's quite complicated, if
a lot of "state" is involved in the application. I was thinking that some
kind of Finite State engine would be quite nice. Does anybody uses something
like this? I'm not sure, but I could imagine, that this way it might be
easier to track changes than using deeply nested if...elif...else
3. Somehow I ended up using my own DB wrapper for MySQL. A few weeks ago we
had this little discussion about element and class ids. I think, in the end,
Chuck thought about to untangle the merged id (32bit class id and 32 bit
element id out of a 64 bit number). Having thought about this a lot, I
really think, to leave it, as it is and give every element this id as the
uniquie id. In this way, every element (row) in the database has a
databasewide unique id and it is possible to extract the class information
out of this id.
The reason for this is the following: I'm still not sure, if it is possible,
to model a m:n relationship with middlekit. In my thin DB Layer, I moved all
relations (1:n and m:n) into a "relation" table with just 2 columns. The
first column holds some objectid that includes some pointer to a foreign
object that is referenced by the second column.
With this, user1 would be member of group1 and group2, user2 memeber of
group1 and user3 member of no group at all.
Since the ids are unique, we wouldn't need special tables for different kind
of m:n relations and, actually, we don't have to differentiate between one
foreign key and and a list of them.
I'd be happy to get some feedback and clarify things, if I didn't make
Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com.
At 09:02 PM 3/11/2001 +0000, Stephan Diehl wrote:
>1. Since Webware has a session management, it might happen, that one is
>visiting the same url twice, but in different contexts (maybe one has an
>application, that collects different data and in the end one goes to an
>url "myApp?action=save". Anyway, if the Browser has caching enabled,
>chances are, that the browser won't even try to connect to the server
>since he already "knows" the url.
>I was thinking, that we needed the possibility to add a "random" attribute
>to every request. This could serve two purposes: a) every url were unique
>and b) one could decide very easily, if a request was valid in my web
I don't know if anyone responded to this yet or not. I'm cleaning out my
mailbox and found this still-unread message.
My first impression is that you would be better off using one the expires:
or cache: type HTTP headers. Otherwise, browsers and proxy servers will
attempt to cache all these pages that really shouldn't be cached. And this
technique still allows you to keep clean URLs.