I have to very much agree with what Robert says below. This year has
been incredibly frustrating (apart from meeting Chuck and a whole lot
of you other great people at WWDC) and part of it is the movement in
the WO space. OK, we had a large JC project which we now have someone
doing it in Cocoa (stuff the Windows users, we only sell Mac anyway),
although I don't know how that is going to work with multi-threaded
large-scale DB access, which EOF gives you for free.
I got keen on Eclipse, the enthusiasm of Chuck, Mike, and Anjo rubbing
off, and I'm sure it's very productive for them. Mike and other's work
on WOLips is really great and I can see how that is coming along, but
I'd have to say that Eclipse still leaves me, well not quite cold, but
luke warm. However, there is no choice but to go with Eclipse, since
Apple really isn't bothering with Java development in Xcode because
the tools like Eclipse and NetBeans specifically target Java, even if
not in a very Mac way. Maybe as Mike says the stable WOLips 3.3.2 will
be out on Friday (that's Monday for us). OK, I think we can blame the
instability of Eclipse 3.3 for some of the months between WWDC and now.
I also agree with Robert that HTML generation tools such as WOBuilder
are an anachronism now. In the current environment, you have to hold
your nose and do the HTML yourself since you have to do CSS,
JavaScript, and AJAX. I'd always held out against doing straight HTML,
because it's just tedious and didn't seem anything a reasonable
graphical builder should not be able to do. So not only has WOBuilder
not kept up with the latest HTML advances, but these other things as
well. I'm sure someone could come up with some kind of graphic builder
for these, but I can't see one on the horizon.
So I have spent a lot of the year, trying to work out what is the way
out of this technological hole. I have researched Java EE (was J2EE),
but found that all so horrible with Hibernate, Struts, Spring, and a
lot of other frameworks trying to get it right and build it into WO/
EOF. That path looks like a definite path to pain, notwithstanding the
1,500 page "tutorial".
Then I looked at Ruby on Rails. I have to agree with Robert that this
is one elegant beast and is fun (whereas the more I look at J EE, the
more I am filled with dread), but is probably not quite in the same
heavy-duty league as WO/EOF (or J EE, yuk). Maybe it will take over
and do the 80% of sites where you don't need anything heavy duty. I'd
still say that Eiffel is a more elegant language than Ruby, although I
think Ruby was somewhat influenced by Eiffel, just as C++ and Java
have been in perverse ways. However, Eiffel is not the flavour of the
month, especially on the Mac, because it is statically typed, and
Objective-C people and now Ruby people like dynamic run-time typing. I
can see the point, but RTT is more the exception than the norm, and
can be catered to in a more static framework (just one or two language
constructs are needed to tell the compiler to circumvent the check).
What I find frustrating about Ruby development, is it is not compiled
at all, and so you have to test by running it, even to find simple
syntax errors, let alone semantic bugs. You have to think about
arranging your tests so that everything gets tested, whereas in a
compiled language, the compiler tests everything (especially in Eiffel
where Make, compilation, and cross-module link consistency checks are
done in the compiler). For that reason, I found Eiffel the best
refactoring language ever. Refactoring in dynamic environments is much
more dangerous. Not only do you have static type safety in Eiffel, but
design by contract which does much of the run-time testing for you as
well as providing formal documentation (much better than JavaDoc and
its ilk, what is all that pseudo-HTML and @ directive garbage). Sure
you still need some Unit testing in Eiffel, but the effort to produce
it is greatly reduced, as is your dependency on testing (which
according to Deming is a good thing). Everything goes in cycles and
one day the world will realize the advantage of the static testing
approach instead of the bulldozer "let's write a test for everything"
approach. When processors are really fast and therefore static builds
more-or-less instantaneous maybe this aversion to building and
compiling will go away. Eiffel does a good job at dependency analysis
rebuilding only that which is necessary, although I do like the way
Eclipse recompiles things automatically in the background. Anyway,
that's a digression into what could be, and why I find where
programming is going at the moment to be frustrating.
The next frustration is Wonder. It really is something that you can't
do without because of its support for the latest Web technologies and
the fact that it patches Apple's bugs and provides a whole lot of
other cool things. But you can't do with it either – it is so
unapproachable to get more than the basic from it. That is one reason
why I'm sure Greg said the R was missing from RAD. I got enthusiastic
about Anjo's BugTracker example at WOWODC – exactly what we wanted for
our D2W reporting project, but then found it so hard to work out how
BugTracker did what it did (along with the Eclipse conversion at the
same time) that eventually all that got canned with the rest of the JC
project to do it in Core Data. I even bought a copy of Professional
WebObjects 5.0 with Java (for $4.95!) to read Max Muller's chapters on
D2W. But that little "learning cliff" turned out too much.
Then there is WO 5.4 fixing some of Apple's bugs obviating the need
for some of Wonder's patches, and putting in some of the latest Web
technologies such as AJAX to a degree, so how Wonder's AJAX plays with
5.4, it seems we don't know yet. And we now have proper support for
generics (in NSArray, etc), although we are still waiting for Java 6.0
(and the Java-dev people seem very annoyed about that since they were,
not really justifiably, expecting it to come with Leopard).
OK, the other reason I'd like to stick to Xcode is that I'm a Cocoa
programmer before a WO programmer (having done Mac since 1984, since
we had one of the first Macs outside of Apple Australia back then when
we cross-developed on a Lisa), and Cocoa programmer since DP 3 in
2000. Now I have to switch mindset in more major ways between WO with
Eclipse and Cocoa with Xcode (character building I'm sure).
So that brings me to the frustration of the Cocoa/Leopard front. On
the plus, Leopard has introduced a lot of developer tools and
frameworks (core animation, QTCapture, etc and much else that we would
have had to write ourselves) that we really wanted back at the
beginning of the year, but had to wait until three weeks ago to get.
The developer builds didn't cut it and since I was doing RubyCocoa,
this did not work with Core Data (a bug I reported and got fixed for
the real release). On the minus was the wait and the instability.
Like I'd say with Eclipse and Wonder, Leopard was something you
couldn't do without, but couldn't work with either. Now that is mostly
fixed and I could start making some good progress, but the PHBs around
here have decided, no they want to drop that and get back to one of
the WO products (the one that doesn't use EOF, so an RoR rewrite looks
attractive especially if we can include SproutCore (but that's not
until January), the one that did use EOF is the one they have decided
to do in Core Data).
I don't want this to sound like an anti-Apple or anti those who work
on WOLips and Wonder rant, because you guys are doing a great job
bringing us the tools we need to keep up with all that is happening in
technology. Apple at least keeps us ahead with its roughly two yearly
updates (although I think Leopard was too big a jump so would rather
see 18 months), which beats Microsoft's pathetic huge jump every five
or six years.
Well, this year has all been a bit like nailing jelly to the wall, but
maybe 2008 will be better, now that we have a reasonably solid
Leopard, the WO/Eclipse world seems to be finally becoming more
stable, and maybe Wonder will be better documented and this stuff will
get to be easier to get into again, rather than approaching the
complexity of J EE.
Ian
On 15/11/2007, at 10:38 AM, Robert Walker wrote:
Greg,
Not to belittle all the work people are going through to make this
work but it seems the R in this RAD product is now gone.
Most of us are finding that it is even _more_ rapid.
I have to say I'm in complete agreement with Chuck on this one. As
many on this list know, I was also hit pretty hard with the
deprecation of the tools to many reasons outside my control. Here is
my two cents on this matter. WOBuilder separated you so much from
the underlying HTML that it literally hurt. For me it was an excuse
not to learn the fine intricacies of HTML. It turns out that's
actually a bad thing. Letting a tool build HTML generally makes for
bad HTML. I used to rely heavily on visual design tools to keep from
getting my hands dirty with HTML.
While it may be great for Desktop Publishing application to
completely hide the formatting and structure code that make up a
page layout. It turns out be problematic for HTML visual editors to
do that.
2. Has anyone migrated to a completely different application
server/development environment and what do they like better about
it?
I doubt they'd be on this list if they did... But I have never
heard any such story. RoR is nice for certain classes of
applications.
Well there's at least one of us still around on the list. The only
time I touch WO now is to maintain our currently installed
applications. This was not, however, by choice. Some of us,
unfortunately, work for someone else, and don't have decision making
authority. We get stuck with whatever the company decides to throw
at us. In our case that's Oracle JDeveloper (J2EE with Oracle ADF).
Oh, how I miss WO....
Now that's at work...
For my personal projects, I have migrated to an entirely different
set of tools. And yes that tool is RoR. I have to say, I LOVE Ruby
on Rails. It is fantastically fun. It has put the joy back into
programming for me. Ruby is the most beautiful language I've ever
had the privilege of working with. The Rails framework is clean,
simple and rich. Would I choose RoR for any application? Probably
not. But, I do choose it where it makes sense.
WO has the guts to play with the big dogs, and EOF is as beautiful
in design as anything in Rails. The real gem, sorry for the intended
pun, of Rails is Ruby. And after working with J2EE for some time
now. I have to say that I took EOF so much for granted. It was the
first Object Relational Mapping tool I learned. I've seen nothing in
J2EE that can hold a candle to EOF.
So if I lived in a dream world, where I actually got to choose my
own tools, that's a pretty easy choice. For large scale, enterprise
applications, WebObjects wins hands down. For everything else RoR is
a wonderful choice, and that warm fuzzy feeling you get when you
write one simple, and elegant, line of code that takes an entire
Java class to replicate...well that's just a nice bonus.
Use the tool that fits the job. WebObjects can handle anything that
J2EE can, while being much more bearable. RoR for the lighter stuff
and just to have some fun. However, underestimate RoR at your peril.
There's some real world power under that hood.
I say give it a chance man. Embrace change! You just might come out
the other side with a simile on your face, and kick ass application
to give to your client.
On Nov 14, 2007, at 1:24 PM, Chuck Hill wrote:
On Nov 14, 2007, at 9:57 AM, Greg Smith wrote:
I have been using WO since the pre OSX client days and have been
loving it. As a heavy WOBuilder and EOModeler user I am finding
WOLips quite a downgrade.
As a heavy WOLips user, I would find going back to WOBuilder and
EOModeler quite a downgrade. EOModeler especially. Which version
of WOLips and Eclipse are you using?
Not to belittle all the work people are going through to make this
work but it seems the R in this RAD product is now gone.
Most of us are finding that it is even _more_ rapid.
It seems this is what set WO apart from the rest, power and ease,
the Apple mantra.
Now for a couple questions for my fellow WO users:
1. Is there a migration process from 5.2 to WOLips or am I stuck
with rebuilding every product by cutting and pasting source?
http://wiki.objectstyle.org/confluence/display/WOL/Migrating+from+XCode+to+WOLips
2. Has anyone migrated to a completely different application
server/development environment and what do they like better about
it?
I doubt they'd be on this list if they did... But I have never
heard any such story. RoR is nice for certain classes of
applications.
3. Has anyone tried something else and gone back to (or wish they
could) WOLips, and why?
Several have. Try J2EE for a few months. :-P
Chuck
--
Practical WebObjects - for developers who want to increase their
overall knowledge of WebObjects or who are trying to solve specific
problems.
http://www.global-village.net/products/practical_webobjects
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list ([email protected])
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/webobjects-dev/robert.walker%40bennettig.com
This email sent to [EMAIL PROTECTED]
--
Robert Walker
[EMAIL PROTECTED]
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list ([email protected])
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/webobjects-dev/ian.joyner%40sportstec.com
This email sent to [EMAIL PROTECTED]
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list ([email protected])
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/webobjects-dev/archive%40mail-archive.com
This email sent to [EMAIL PROTECTED]
