On 12/13/2010 10:55 AM, Ben Imp wrote:
> Ah yes, not enough animal sacrifice in my deployment procedures.  That
> must be a significant part of my problems.
> 
> I have dealt with this issue before in standard Java RMI based client/
> server applications.  However, in those situations, we had a more
> capable communication pipe between the client and server.  There it
> was possible for the server to initiate communications, and slap
> around the client if it the users got all uppity and refused to close
> it.  This made deployments rather easy, and stroked our god-complex-
> driven egos to boot.  In GWT (or just the web in general, really), we
> don't quite have this.  At least not without some hacky client
> polling, which I find just slightly less than elegant.
> 
> I do agree with the point about not really being able to provide a
> common solution to the issue.  It would be a bit odd if GWT locked us
> into a more rigid communication structure in order to auto-magically
> handle version differences.  Additionally, having the framework force
> a browser refresh seems a bit evil, and I don't like evil, as a
> general rule.

GWT's reliance on browser refresh has to satisfy two competing
objectives: caching and on-line updates.

It's also the case that nebulous documentation exacerbates the tension
between those objectives.

The informal nature of much on-line documentation encourages the failure
mode that is this thread's subject. Is the expectation that we now
mashup N blogs, wikis, javadoc et. al. just to achieve parity with
professionally created documentation? Software's rapid development pace
makes a farce of documentation's craft, yet it's that parody upon which
we now rely.

> 
> Now I just need to find a cow to murder for the next release ...
> 
> GWT: Its delicious beef-flavored development.
> 
> -Ben
> 
> On Dec 13, 10:38 am, Jeff Chimene <[email protected]> wrote:
>> On 12/13/2010 08:52 AM, Ben Imp wrote:
>>
>>
>>
>>> Excellent - I am not alone in my troubles.  On second thought, perhaps
>>> rejoicing in the sharing of pain is not the most healthy viewpoint to
>>> take ...
>>
>>> The post to the older discussion was quite useful.  The blog linked
>>> there has some pretty good information on how this same issue has been
>>> dealt with in the past, and some other issues that I haven't yet had
>>> to deal with.  In the end, it seems that it all comes down to crafting
>>> your application in such a way that it is able to perceive its out-of-
>>> dated-ness (English is awesome) and inform the user of the potential
>>> errors, or just force them to do the Right Thing and reload the
>>> application.  The latter seems like a more sane thing to do, given the
>>> similarity of instructing users to herding cats.  I suppose false
>>> positives are something that might be in issue in that case, and an
>>> epileptic application may result, but that should be easy enough to
>>> discover in testing.
>>
>>> I agree that this should be mentioned in the documentation as well.
>>> It seems like something that is much easier to deal with if you
>>> consider it from the beginnings of your application, as opposed to
>>> running into it (or, perhaps, more appropriately, stumbling into it)
>>> somewhere down the line.  One would assume the Google guys (aka Sir
>>> Developer Not Appearing In This Thread) have dealt with this, since
>>> their AdWords UI is (if I recall correctly) based on GWT.  I would
>>> likewise assume that product has a fair number of active users at all
>>> times of day, and they are probably as hesitant to drive around the
>>> world as I am.
>>
>>> I would be most intrigued to see what they have done internally.  I
>>> imagine it involves pixie dust and invisible pink unicorns.
>>
>> And slaughtering a goat on the keyboard.
>>
>> ISTR that this issue is similar to the authentication/authorization
>> issue in that there exist many implementation-specific constraints. GWT
>> gives you enough information to make a run-time decision about
>> out-of-dated-ness.
>>
>> To be fair, this issue surfaces in any long-running client-server
>> application suite (e.g. COBOL/IMS)
> 

-- 
You received this message because you are subscribed to the Google Groups 
"Google Web Toolkit" group.
To post to this group, send email to [email protected].
To unsubscribe from this group, send email to 
[email protected].
For more options, visit this group at 
http://groups.google.com/group/google-web-toolkit?hl=en.

Reply via email to