A turbo class vapouriser!

[ConvertFromOldNGs]

by Craig Shearer >> Tue, 21 Mar 2000 22:50:47 GMT

Adding new classes in JADE is easy. But, once you've added a few, and connected them together with countless references, changing your mind is very difficult. Usually, to delete a set of classes you're created for "experimental" purposes, you need to delete the classes in a very specific order, first detaching any references, removing collection classes (for those that have keys using properties of other classes) etc.

What JADE really needs is a "turbo class vapouriser" to allow the developer to zap a class or set of classes. Of course, there'd have to be warnings etc. about the severity of the action, but JADE should then just do it, and get rid of whatever it needs to. I suggest that a summary dialog appears that would show you the consequences of your actions.

What do others think?
--
Craig Shearer
Email craig.shearer@bigfoot.com

[ConvertFromOldNGs]

by Darrell Duniam >> Tue, 21 Mar 2000 22:54:45 GMT

Perhaps a slightly more elegant way would be something akin to an uninstaller, where you are shown all of the related bits and pieces, and you can tick which ones you want to remove then do it all in one go.

darrell.

[ConvertFromOldNGs]

by Robert Barr >> Wed, 22 Mar 2000 1:01:57 GMT

When you get a "references exist" message, could JADE then ask "do you want to delete the references too?".

(Just as a reminder - quick way to remove instances of "experimental" class structures is to delete the map file - this can save a JadeScript or three).

[ConvertFromOldNGs]

by Carl Ranson >> Wed, 22 Mar 2000 2:47:53 GMT

Correction, Craig.

There should be precicly ONE warning, summarizing the extent of the requested action. Having multiple warnings for one action is idiotic. Its merly a cop-out to make the developers life easier.

If I want to delete a class, I should be allowed to. Jade should "help" me by advising the effect of my actions, but it shouldn't then stop me, unless its going to be fatally destructive.

Remember whos in charge here, people. We tell Jade what to do, it shouldn't presume to tell us.

I challange anyone to give me a valid interaction design reason for having multiple warnings.

Regards,
CR

[ConvertFromOldNGs]

by Craig Shearer >> Wed, 22 Mar 2000 21:26:06 GMT

Carl

I agree with your sentiments, however, perhaps you misunderstood my intent. Multiple warnings do not necessarily imply multiple dialog boxes - that's just the default JADE implementation - the developers being lazy and deciding that the Windows Message Box is good enough for any purpose.

I was meaning that there'd some sort of summary dialog explaining all the consequences of the action - one dialog, not 10!

Craig.

[ConvertFromOldNGs]

by Gregory Tolhurst >> Wed, 22 Mar 2000 22:21:02 GMT

Carl

I agree with your sentiments, however, perhaps you misunderstood my intent. Multiple warnings do not necessarily imply multiple dialog boxes - that's just the default JADE implementation - the developers being lazy and deciding that the Windows Message Box is good enough for any purpose.

It's awfully easy to find developers guilty of laziness when their behaviour is more likely to be motivated by cost-effectiveness.

When I get to work in the morning and I have to open the door of the office building then the door to my wing then the door to my office, I don't blame the architect or builder for laziness. I figure that a direct route from where I park my Skoda to my desk would cost more than it would deliver in value.

Frederick Brooks (The Mythical Man-Month) observed that the tractability of the software medium leads people to expect all their desires to be met by software developers. I'm sure that Craig wouldn't expect his employer to spend money on timber/concrete/steel to get an external door beside his desk. Why does he think that the Jade developers (who are paid more than carpenters) should select the more expensive way of solving an interaction problem?

Gregory

[ConvertFromOldNGs]

by Craig Shearer >> Thu, 23 Mar 2000 0:33:49 GMT

It's awfully easy to find developers guilty of laziness when their behaviour is more likely to be motivated by cost-effectiveness.

When I get to work in the morning and I have to open the door of the office building then the door to my wing then the door to my office, I don't blame the architect or builder for laziness. I figure that a direct route from where I park my Skoda to my desk would cost more than it would deliver in value.

Frederick Brooks (The Mythical Man-Month) observed that the tractability of the software medium leads people to expect all their desires to be met by software developers. I'm sure that Craig wouldn't expect his employer to spend money on timber/concrete/steel to get an external door beside his desk. Why does he think that the Jade developers (who are paid more than carpenters) should select the more expensive way of solving an interaction problem?

Gregory

A number of points here:

1. You are assuming that it does cost significantly more to develop an elegant solution (say one form with multiple messages on it) than one that uses the Windows Message Box. I don't believe this to be true. JADE could have a reusable form that could display multiple messages.

2. More importantly, to provide competitive advantage. I would much rather use a software development tool that had a usable interface for developers than one that didn't. There are other development tools that treat the developer much better than JADE does. JADE is a powerful tool for developing systems, but it is hampered by the user-interface it presents to the developer.

3. I don't think that the analogy is appropriate. Software can be much more accommodating to individual user preferences than buildings can. That's why it's called "soft"! I do agree with Fred Brook's comment though - but that's reality. If one software product doesn't satisfy users' desires, then users will often be willing to switch to another one that does.

Craig.

[ConvertFromOldNGs]

by Gregory Tolhurst >> Thu, 23 Mar 2000 22:04:33 GMT

1. You are assuming that it does cost significantly more to develop an elegant solution (say one form with multiple messages on it) than one that uses the Windows Message Box. I don't believe this to be true. JADE could have a reusable form that could display multiple messages.
<snip>
3. I don't think that the analogy is appropriate. Software can be much more accommodating to individual user preferences than buildings can. That's why it's called "soft"! I do agree with Fred Brook's comment though - but that's reality. If one software product doesn't satisfy users' desires, then users will often be willing to switch to another one that does.

Aren't words wonderful in the right hands? If I argue for a message box solution, then Craig has me arguing against an "elegant" solution!

Yes, it's true that a reusable form displaying multiple messages could be inexpensive to develop in itself. Implementing it could be very expensive though...

Putting up a message "You can't delete this class because references to it exist" is easy (and inexpensive) for the product developer (some would say "lazy"!).

Putting up a message "If you delete this class the following references will also be deleted..." is harder (and more expensive) because the product developer now has to design another interaction. He has to write a readable message that may extend for hundreds of words. He has to provide more options (continue and delete all references; continue and unhook references in case they embrace a superclass of the class being deleted...). He has to handle support issues from customers who say "I got a thirty-line message so I just clicked OK and now all my references have vanished".

Analogies are like farting in church - seldom appropriate but hard to resist. Oops, there's another lame one! Yes, my building analogy is weak, but my point remains: features aren't omitted out of laziness - they're omitted for more complex reasons.

Gregory