RE: No 5250-based applications -- MIDRANGE-L



>
> Exactly the same can be said if you remove the back button, especially if
> you Javascript around ALT-leftarrow, which I'm pretty sure you
> can do.  But
> even if you can't, ALT-leftarrow is NOT an accidental keystroke.
> Provide a
> default, non-destructive action and move on.  You really need to
> lighten up
> on this issue.

That made my chuckle... you've obviously received the impression that I have
a bee in my bonnet about this - I don't. I'm just discussing it. Entire
articles have been written about this "problem".

http://www.4guysfromrolla.com/webtech/111500-1.shtml
http://www.centurasoft.com/support/products/cwd/white/sharecwd.asp

A websearch returns countless pages with quotes like the following:

"A common problem most web application developers will encounter is
the one a browser's "back" button introduces to certain processes. The
ability of a user to jump back in the middle of a critical logical
process can be both disruptive and dangerous if not handled correctly."

Me thinks I'm not alone in my concern!

> Lemme know when you're done scratching.  I long ago decided the advantages
> of a completely platform-independent interface that required absolutely no
> additional workstation code and could access my systems from
> anywhere there
> was an Internet connection FAR outweighed any personal quirks I might have
> about not wanting to code around the back button.


Platform independent? Not if you consider your platform to be the various,
and incompatible, web browsers. So instead of tweaking for various
platforms, we're now tweaking for various browsers. Six of one, half dozen
of the other.

As far as having additional workstation code - who cares if there is. We
have technology that allows automated software updates over the web, so it's
not like you need to run around doing a bunch of manual installs.

Most of the internet software that I have calls home to check for updates as
soon as it starts up. If any are found, it installs them automatically, and
carries on.


> What the heck does F12=Previous have to do with any of the previous
> discussions?  F12 is NOT a back button, unless you also have F13=Forward,
> wherein you go to whatever panel you just came from, with all your data
> filled in.  And I am absolutely positive you don't have that.

This issue is navigating between panels of an application while staying in
context. With a browser, it's very easy to get the user interface in a state
where it's no longer synchronized with the application. Apparently, it
doesn't seem to be very easy to prevent that from occurring.


> On the other hand, every one of MY revitalized applications enables every
> command key of the original applications with a simple button, so
> in effect
> I have a "Back" button (or more correctly, a "Previous" button), if the
> original application had a "Previous" command key.

Put 'em up on the web. I'd love to see how they work.


> Caching at the browser?  Stateful connections?  Huh?  At one client, I am
> currently using the browser as a complete green-screen replacement - it
> supports subfiles and field attributes, just like on your 5250 screen.  At
> another client, I am using a completely stateless interface to transaction
> servers, but again taking advantage of HTML to emulate attributes such as
> field protection and error notification.

You make it sound like you're doing something radically different than the
status quo in web development. If you are, and it's an improvement, I'm sure
the world will be beating a path to your door. I for one, will be the first
to say "thank you Joe!".

> The browser can be used for either stateful or stateless connections - it
> just requires a good UI architecture.  Not only that, with my
> architecture,
> I can just as easily replace the browser with a Swing thick client to
> provide a more "solid" integrated feel, as opposed to the browser.  And in
> THAT environment, I don't have to worry about somebody hitting
> ALT-leftarrow.

I have no doubts about the thick client doing the job. That's the way I do
it, so I know it works. The browser on the other hand, by virtue of the fact
that it uses a stateless protocal (HTTP), is by definition a stateless
connection. Or so I thought... but I'm still learning. :)



John Taylor
Canada

+---
| This is the Midrange System Mailing List!
| To submit a new message, send your mail to MIDRANGE-L@midrange.com.
| To subscribe to this list send email to MIDRANGE-L-SUB@midrange.com.
| To unsubscribe from this list send email to MIDRANGE-L-UNSUB@midrange.com.
| Questions should be directed to the list owner/operator: david@midrange.com
+---