Ascend Archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
(ASCEND) Another round
I hate to bring this thread up again, but it does bear on the stability of
any software including Ascend's OS.
Who are Ascend's customers? And do we really need all these features?
I've got this TNT with "features" I can't even begin to guess what they
would be used for and suspect they aren't used by 99% of the people who own
them. But I have no doubt whatsoever that they negatively impact the
stability (and performance) of the OS.
I'm reminded of IBM's OS 360 project (does that date me:). This was a
product that was too ambitious for the design, organization, coding and
testing methods of the time. As a result, 60% of the features never did
work. And every time one bug was fixed it would let you into a new section
of code where 10 more bugs would be discovered. Finally, IBM gave up
trying to fix it and froze the release because the harder they worked the
longer the bug list and customer complaints got.
While design methods, programming methods and languages have done a lot to
address large project issues, every method has its limit with respect to
complexity. More complex projects must wait until better methods are
developed.
To increase stability may require a reduction in complexity. That is, if
Ascend's OS is to ever become truly stable, there may need to be a
reduction in complexity at this stage of the project.
So again I ask, who are Ascend's customers? What features do they not
need? What features are they willing to give up for increased stability?
Apple Computer showed the world what a mistake it was to ignore their
customer base and go chasing after a market share that didn't want them.
Ascend is chasing RBOCs. Are they a significant part of Ascend's market?
Does Ascend believe the myth about ISP consolidation and RBOC dominance?
When Netscape first came out with their servers for Windows NT, I ran them
for nearly a year before changing to Microsoft's somewhat more modest IIS
1.0. There was only one issue involved in my decision - stability.
Netscape's server crashed every three days and Microsoft's server did not -
one went 15 months without a crash or reboot. In a production environment
with customers depending on you, stability wins over "features" everyday of
the week.
I would like to see this list die for lack of interest because every Ascend
box out there is humming along perfectly.
--Tony
----------
> From: Kevin Smith <kevin@ascend.com>
> To: Jason Nealis <nealis@babylon.erols.com>
> Cc: Ascend User-Group <ascend-users@bungi.com>
> Subject: Re: (ASCEND) 5.0Ap34
> Date: Tuesday, November 25, 1997 10:16 AM
>
> At 09:52 PM 11/24/97 -0500, Jason Nealis wrote:
> >On Mon, 24 Nov 1997, Jorge Paramo wrote:
> >
> >> On Mon, 24 Nov 1997, Jason Nealis wrote:
> >>
> >> > test. If they pushed everything through full regression testing then
you
> >> > would only see a patch release once a month. The main focus of
> >>
> >> Yes, but that would make those patches a LOT more effective than the
ones
> >> they have been releasing lately...
> >
> >Correct, But then the topic of this flame would probably become, Why
> >does it take ascend a month to release a patch release to fix a
> >certain bug. etc..
>
> Were you around when _that_ was the complaint ?
>
> There are many answers to this problem, as you can see from this thread,
> everyone has different ideas - we all have the same goal.
>
> We *did* have longer release times in the past, and didn't make patches
> available (except to targeted customers who had to sign beta agreement
forms).
>
> That was not acceptable either...so we made those BETA releases available
to
> everyone with a few README disclaimers....they were sometimes too
unstable...
> so we got different complaints...
>
> Then we split the branch to patch and incrementals....trying to keep only
> bug-fixes in the patch branch and new features only in the incrementals
> branch.
>
> ONCE AGAIN that was not good enough - we were asked to add new features
to
> "stable" patch releases - so now we *occasionally* add enhancements to
patch
> releases. More complaints? - You bet !
>
> We are trying to do the best to satisfy the largest number of customers,
so
> some
> will not agree on our methods or approach, we cannot please everyone from
what
> I've seen so far (and I've been doing this now since the VERY early days
of
> Ascend - 7 years ago).
>
> As I mentioned a few weeks ago, we are working on a more methodical
> approach to
> software releases - all taking into consideration what you guys have been
> telling us. The releases will likely come out more staggered - so the
time-to-
> market complaints will likely be back, but the level of stability will
> hopefully
> be increased in both patch and feature releases.
>
> I suspect that we will announce/make available the details when 6.0 is
> released,
> but I have no firm date for when that will be yet.
>
>
> Kevin
>
>
> ++ Ascend Users Mailing List ++
> To unsubscribe: send unsubscribe to ascend-users-request@bungi.com
> To get FAQ'd: <http://www.nealis.net/ascend/faq>
++ Ascend Users Mailing List ++
To unsubscribe: send unsubscribe to ascend-users-request@bungi.com
To get FAQ'd: <http://www.nealis.net/ascend/faq>