Re: An Ada IDE and discussions

["chris.danx" <chris.danx@ntlworld.com>]

"Brian Orpin" <abuse@borpin.co.uk> wrote in message
news:f49tktghbv140bg63a292jul1h940pj5pt@4ax.com...
> On Thu, 12 Jul 2001 02:47:46 GMT, "DuckE" <nospam_steved94@home.com>
> wrote:
>
> >So yes, I do see the need for these features in the form of an IDE for
> >Windows.  The Windows front end on Emacs is not consistant with other
> >windows programs.  Such an editor is definitely desirable.
>
> Your assumption is that all users are 'Windows' users.

That's not what Steve(?) said at all.  "an IDE for Windows" is not saying all
users are Windows users, it's an IDE for Windows users.

If ppl want Ada to become popular, you have to appeal to the biggest potential
user base, which is the Windows platform.  However much some of you may dislike
the OS, it is the domininant OS and there is no denying it.

I'm doing this IDE because

a) I (and others) need compiler configurability

b) It's an interesting problem comprising of many different
   problems, algorithms and solutions

c) It gives me experience with working on slightly
   bigger apps and managing it all

d) It's a project I can do

e) I'll learn from it.

f) It may do something for Ada*

g) It'll be free and it's


[   Oh yeah, no way am I distribing patches to versions in dif(f) files.  I hate
them, I have nothing but trouble with them, and see no reason for their use in
the modern age.  Everything you need will go in one zip and no bloomin'
iffs.  ]

Jerry Coffin and I have had a discussion of various issues related to the IDE
and it's design on the alt.os.development thread "Re: Ada ? C ? was Re: Jochen
Liedtke has died".  I know the title is a bit iffy, discussing Ada under a
thread that started with a principle L4 developer, but you may want to take a
look through the discussion if curious about the IDE and where it's at...


> Emacs predates

No argument

> Windoze and is available on just about any platform and any OS.

I see no reason why that should be achievable for an IDE (though I have no wish
to do so at present).  One possibility would be to port gwindows to wine or
something, but i'm unsure about this and it's a lot of work.

> You can
> invoke modes to make it 'more Windows like' if you wish.

Yeah, but can you click..click..click , and do this if you don't know how to use
it?  I've never had to look through a bunch of help files to find out how to use
anything like JBuilder or ConTEXT!

[** HCI bit **]

Fitt's law:

(try http://ei.cs.vt.edu/~cs5724/g1/#procons for examples of fitt's law)

tell's us that keystroke interface will win over a totally GUI based solution,
for competent users.  But GUI interfaces do feature keystrokes for at least the
critical functions, making them as fast for competent users for critical tasks.
For non-critical tasks, there is a penalty for the traditional GUI interface.


My thoughts:

The penalty for GUI systems is offset by the familiarity of the interface and
the previous knowledge the user has accumulated in the world (other GUI IDEs).
A keystroke interface, while desirable for experienced users, doesn't allow
knowledge accumulated in the world to be transferred as readily to new
scenarios.  In the case of Emacs however, this is eliminated by the near total
uniformity of the key interface accross various different editing modes.

Comparing the learning curves of a GUI IDE, and a keystroke interface leads me
to the conclusion that a GUI IDE is a suitable candidate for the majority of
uses in circumstances where the time available is tight, or the developer is
eager to get working and they are unfamiliar with the Emacs interface.

Also this analysis applies to the more traditional GUI technologies e.g. linear
menus.  A pie menu on the other hand may offer comparible performance to
keystroke interface, though this needs further investigation.  The benefit of
the pie-menu is negated to some extent by the time it will take for a user to
learn this concept, and become familiar with the mouse ahead sequences.  The
latter will IMO require less time to learn than the keystroke approach.  While
the former requires very little time (and at least a two button mouse :-) ).


NOTE:

If the developer is familiar with Emacs, then he or she is free to use it,
unless their manager objects.

[end HCI bit]


> That is of course what Tutorials (built in) and FAQs are about.

Again, the Windows GUI based IDEs are much easier to learn to use (ok, there are
some oddballs, but those designers haven't read "the design of everyday things"
by Norman :-) ), and it's likely
managers don't want to have a high learning curve for tools.


Chris Campbell

*Ada 95 is a nice language but I'm no die hard.  It's not a good tool for
certain problems and I freely acknowledge this.  When a problem arises and Ada
doesn't offer an elegant solution other languages are used.