The only thing that I've studied as much as WebObjects is the Bible,
and that is sometimes easier to explain than WO!
Seriously, the concepts are so deep that I think we (WO developers)
kind of take many of the core concepts for granted and sort of treat
them like riding a bike. Working with them is easy once you work with
them, but difficult to explain up front. WOD files, framework
operation, and EO models are pretty easy to describe, but how, without
having the patience to read the manual, do you really come to
understand the editing context - object store "stack?". Really, how
many people still lock ecs wrong?! How many people screw up binding
handoffs in the R-R loop? Once you move on past the satisfication of
the Hello World app and get your first binding snafu when you change a
bound array in the R-R loop, you can become incredibly mystified and
then frustrated. What about the first time your app mysteriously locks
up under load because you didn't lock an ec right?
If you have the patience to get past these and learn the right way,
then you have to familiarize yourself with all the marvelous things in
Wonder, which is where you should start anyways. It is the only thing
that fixes so many stupid limitations with WO as distributed, and
gives you the ability to do all the cool stuff you want to do anyways;
how can you write any app without Ajax now? Too often I feel like
Wonder is treated as an add-on people get around to when they finally
need it. By that time they've wasted a bunch of time rewriting things
that the Wonder guys have done already. It should be a foundation for
app development, not an add-on. You can read the whole API, and look
at the examples, but I think that for casual users coming from
Rails, .net, PHP even, that is a pretty big barrier to entry.
Perhaps the way forward isn't to document every single API, but to
produce a new "master template" for new users. I could see an example
app (ie HelloWorld) that uses Ajax, nested ecs with locking, ERIndex,
AjaxLongResponsePage, PDF, er.plot (or google chart), er.excel,
er.attachment, er.migration, etc. - whatever we deem as gems.
This skips the learning part (which is still the only way to really
move forward IMHO), but at least with one app new users could see all
kinds of stuff. I know D2W already does a lot of magical things, but
for developers of "traditional" apps, this might show WO flexing its
muscles. This could be cobbled together from most of the examples
built already.
Looking at WO as distributed can elicit a lot of laughter; it is like
an artist showing off their paint and brush collection. With Wonder
you have a painting. It takes the basics and makes WO act like a
modern development framework. I am stuck at 5.3, but lack very little
because Wonder is giving me generics, etc. (I know, I know, I should
upgrade anyway)
If there is another problem, it is in there not being a set of best
practices. For instances, should we assume that everyone is using
ERKey to make qualifiers, or is the instruction to make qualifiers by
hand? Many frameworks take that choice away from you and tell you how
to do it, which in some ways makes it easier to teach. In order to
really grow the base we'd have to teach people to use eogenerator and
ERKey, but keep awareness of EOQualifier in there to maintain
flexibility if they want it, for example. That would be the hard part
and one that all involved would have to coordinate on.
So that's my bit. In the past two months I've been bit by a stupid
NotificationCenter oversight (me = stupid) and missed a big efficiency
in AjaxResponse, so I am hardly an expert. But that is the point. The
people we want to attract aren't either. At least I have the
confidence to know that the answer is out there. Newcomers might not
persevere as much.
John
John A. Larson
President
Precision Instruments, Inc.
Ph: 847-824-4194
Fax: 866-240-7104
Sent from my iPhone
On Oct 21, 2009, at 4:06 AM, Jeremy DE ROYER <jeremy.dero...@ingencys.net
> wrote:
Hello,
Results are interesting but unfortunately they show that webobjects
technologie miss beginners...
Jérémy DE ROYER
Le 20 oct. 09 à 21:01, Pascal Robert a écrit :
Ok, so I bugged everyone for weeks and I don't expect any more
answers (even if I know that at least 5 organizations didn't
respond...), so I made a summary of all answers we have so far :
http://wiki.objectstyle.org/confluence/x/doBf
http://wiki.objectstyle.org/confluence/x/E4Bf
The surveys are still open, I will close them mid-November.
https://www.survs.com/survey/T8TGXAW70R
https://www.survs.com/survey/3O47WI4G11
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (Webobjects-dev@lists.apple.com)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/webobjects-dev/jeremy.deroyer%40ingencys.net
This email sent to jeremy.dero...@ingencys.net
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (Webobjects-dev@lists.apple.com)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/webobjects-dev/the_larsons%40mac.com
This email sent to the_lars...@mac.com
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (Webobjects-dev@lists.apple.com)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/webobjects-dev/archive%40mail-archive.com
This email sent to arch...@mail-archive.com
