On 17 Dec, 15:44, Mark Woyna <wo...@[EMAIL PROTECTED]
> wrote:
> On Dec 14, 11:02 am, ap...@[EMAIL PROTECTED]
wrote:
>
> > Occasionally (3 or 4 times times now in the last 3 months) orbixd
> > appears to go into CPU meltdown on our UAT machine. No-one can fathom
> > why. All we can do is restart our processes until the problem goes
> > away, as mysteriously as it arose. I wonder if anyone else has seen
> > anything like this. We are using an ancient, unsup****ted version of
> > OrbixWeb (3.0.x).
>
> > -Andrew Marlow
>
> I guess I'm beating a dead horse here, but I have to question the
> sanity of an organization that continues to suffer with an ancient,
> buggy piece of software, rather than upgrade to something current. The
> number of issues you've encountered are roughly the same number I
> experienced nearly a decade ago. It's time to move on.
Indeed. I never argued that the choices were sane. Allowing the sw to
become so out of date wih the maintenance issues it has IMO is, ahem,
sub-optimal. But I have to make it work until they can migrate away
from CORBA completely.
>
> How is wasting your time saving your organization any money,
> especially when there are plenty of free orbs out there?
It's not my organisation - I'm a contractor. And I did use jacorb to
do most of the development, which was painless by comparison. I did
argue for using jacorb but lost that argument. I also lost the
argument to change to a much more recent version of Orbix even though
we got a VERY reasonable quote from IONA.
> As a general
> rule of thumb, many organizations take the 'if it isn't broke, don't
> fix it' route.
But it was broken (bugs and lack of robustness in the C++ app code).
And using such as ancient unsup****ted buggy mix of technologies they
brought me in the migrate them away from all that. Now the servers
have been rewritten in java and I am in the process of moving them off
CORBA.
> They're not going to throw out working software just
> for the sake of running the latest and greatest. However, it's quite
> clear here that it *is* broken, and it can't be fixed.
Yes it was broken but now the servers are rewritten in java things are
alot better. Now theres just the task of moving away from the CORBA. I
am in the intermediate stage where we have brand new java code and
people need to get that warm comfortable feeling that all the new code
is ok. Part of that is not having the change the clients at the same
time due to a change in the mechanisms for inter-process
communication.
>
> I know it's not your choice. I hope you're learning something valuable
> as you struggle through this.
I'm having a great time. My java has really improved and I am learning
it in the context of java enterprise technologies including JMS. I
have also had a play with swing. And I'm working with people I like.
And I've had the chance to renew some personal ties with the people at
IONA. Yes there have been struggles but its my job to make things
work. I have taken the approach of iterative incremental development
rather than big-bang. That's why there is CORBA for now.
Also, I think that CORBA is a good technology for the problem at hand,
it's the ancient version that is the problem. I did recommend that
they stay with CORBA but was overruled. The fact is that CORBA skills
are rare and expensive something cheaper and more maintainable is
needed in terms of finding the people to look after it. It took them
ages to find me :-)
> I wouldn't be that patient. If my
> organization saddled me with a crappy piece of code like OW, I'd have
> resigned ages ago.
I'm a contractor and will be saying goodbye in February :-)
>
> Mark
|
