Re: Did anyone attend the Webfacing lab at COMMON? -- MIDRANGE-L

>>I came away from the Webfacing sessions with the distinct
>>impression that IBM is here to stay.
>
>I take this to mean that IBM is committed to Webfacing?
>I wonder how committed?

I certainly can't predict IBM's future behaviour, but they've got people
very dedicated to making it work TODAY.  The companies (current users) I
spoke with all said that IBM jumps on bugs and issues patches very quickly.
In addition, they're bundling it with the compilers, so they're essentially
giving it away to development organisations.  That's the carrot...

>> I'm not certain that this is going to work for us in particular
>> (we use Cool:2E and are exploring other possible routes to
>> Internet-enable our applications)...
>
>Cool:2E generates 5250 RPG doesn't it?  Why wouldn't
>Webfacing work for your applications?

Yes, 2E generates RPG and DDS.  Therein lies the problem.  When a developer
makes modifications to the UI part of the Webfaced app, the changes are
stored as comments in the DDS, so the development tools know what was done
and what to do in the future.  2E will CLRPFM the DDS source, wiping out the
record of the modifications.

>> It's NOT screen-scraping, so things like pop-up windows actually
>> open a new window, but like all browser based interfaces, it suffers
>> from "don't back up" syndrome.
>
>5250 emulation doesn't support the <Back> button either.  Is that
>really a problem?

It is for end-users accustomed to browser behaviour.  It can be lessened by
removing the "back" button from the toolbar, but keyboard people like me
will still send the application into the dead zone.

>> Once the Java stuff is running, the programmers will start
>> tinkering, and Web-technology will then become as "iSeries-ish"
>> as RPG is today.
>
>Is this a prediction that most applications will be written in Java?

My crystal ball is as good as anybody's, I guess.  I suspect that we will
see a lot of the RPG "converted" into stored procedures or equivalent in
order to retain the business rules.  Let's face it; most AS/400 companies
have no written record of what the business rules are - how could I possibly
re-write my apps in Java?  All that "ugly" code is there for a reason -
maintenance, bug fixes, etc.  If it's possible to do a relatively small
amount of work to put a graphical face in front of that, it seems like a
bargain!

>> Is there something in particular you were looking for?
>
>Had I been able to attend the conference, the Webfacing
>sessions would have interested me.  Nobody seems to
>know how much more CPU is required to run a Webfaced
>application.  That's one question.

The stick.  Over and over I heard that WebSphere Application Server needs a
machine with as much L2 cache and MHz as one can get.  (WebSphere or some
other servlet engine is needed to serve up the JavaBeans created by
Webfacing.)  I couldn't get anybody to tell me if my 620 will run WebFaced
apps OK, and there are no published benchmarks for older models using this
environment.  All I hear is "buy a new machine."  One glimmer of hope could
be Jakarta Tomcat which is lighter-weight than WAS.  But the reality is that
doing Java presentation work on the iSeries will consume more cycles than
the 5250 "equivalent."

>I once offered the hypothesis that Webfacing was a
>temporary transition technology to encourage
>customers to buy into Websphere.  Is that true?
>Or is the DEVELOP UNDER 5250 then DEPLOY
>VIA WEBFACING a viable long-term development
>strategy?

Excellent questions, Nathan!  I think that WebFacing is targetted to the
smaller customers; folks who don't have Web expertise and don't intend to do
more than a token effort to Internet-enable their apps; probably to
eliminate SDLC costs to remote branches.  They don't really care if the
WebFaced app is ready for public internet consumption - all they want is to
be able to reduce their communications infrastructure costs.  In this
environment, I'd say that WebFacing is indeed a viable long term strategy.
To turn the idea on it's head, try asking if "array-based S/36 displays are
a viable long-term development strategy?"

For vendors like us, I think WebFacing is transitional, designed to whet our
appetite for a JavaBean-based UI.  I suspect that IBM win in this scenario
because they get to sell education on the new development tools and WAS
(services revenue rocks), as well as WAS and hardware sales.  Further, they
benefit because the iSeries begins to shed it's stodgy image.  Application
developers win because they're finally getting Official IBM sanction and
encouragement (low price!) to use Java based technology.  End user companies
win because they get an Internet-enabled app. fairly quickly, as well as
happier developers.
  --buck