From: "Rich" 

This is a multi-part message in MIME format.

------=_NextPart_000_00C2_01C2C87A.BD2FB1A0
Content-Type: text/plain;
        charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

   Not testing the update.  That is constant.  Testing a full install =
after files have changed.  This is something unique to the release of a =
full install.

   Breaking things is not a simple thing like you suggest.  The kind of =
breaks I hear people complain about here are not that the application =
being updated broke as much as some application or process that used it =
did.  That gets tested for but it's impossible to test every possible = app
in every possible configuration.  People are fallible.  What = sometimes
happens is that these applications have dependencies on broken = or
undefined behavior and are in need of their own fixes.

   How you decide to deploy fixes is really up to you.  No one can tell =
you what is best for you.

Rich
=20
  "John Cuccia"  wrote in message =
news:9k2j3v46qc6l7tiqrogk9hvqscboj0bv5c{at}4ax.com...
  On Wed, 29 Jan 2003 22:21:36 -0800, "Rich"  wrote:

  >   For other reasons, I think it is impractical.  You expect a full =
release for any update.  This is a good way to discourage any software =
vendor from making releases.  The testing effort is enormous.

  Testing is enormous in any case: either the vendor does it before
  update/release or *every* conscientious customer does it before
  updating production machines.  Naturally, the selfish vendors prefer
  to offload the testing to customers.

  >  Updates are very easy in comparison.  They are easy to apply and =
take very little time.=20

  Only If they don't break anything.  Since they are not thoroughly
  tested by their creators, customers are forced to either test them in
  a lab environent or to roll the dice by deploying an untested fix to a
  production environment.

  My selfish viewpoint: It is (e.g.)  Microsoft's product, let them test
  test it.  They charge me enough for software without requiring me to
  test updates that do no more than fix holes in the products they sold
  me in the first place.=20




------=_NextPart_000_00C2_01C2C87A.BD2FB1A0
Content-Type: text/html;
        charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable








   Not testing the =
update.  That is=20
constant.  Testing a full install after files have =
changed.  This=20
is something unique to the release of a full install.
 
   Breaking
things is not a =
simple thing=20
like you suggest.  The kind of breaks I hear people complain about
= here are=20
not that the application being updated broke as much as some application = or=20
process that used it did.  That gets tested for but it's
= impossible to=20
test every possible app in every possible configuration.  People = are=20
fallible.  What sometimes happens is that these applications
have=20 dependencies on broken or undefined behavior and are in need of
their = own=20
fixes.
 
   How you
decide to deploy =
fixes is=20
really up to you.  No one can tell you what is best for =
you.
 
Rich
 

  "John Cuccia" <jcuccia{at}bigfoot.com>">mailto:jcuccia{at}bigfoot.com">jcuccia{at}bigfoot.com>
wrote =
in message=20
  news:9k2j3v46qc6=
l7tiqrogk9hvqscboj0bv5c{at}4ax.com...On=20
  Wed, 29 Jan 2003 22:21:36 -0800, "Rich" <{at}>=20
  wrote:>   For other reasons,
I think it is=20
  impractical.  You expect a full release for any update.  =
This is a=20
  good way to discourage any software vendor from making releases.  =
The=20
  testing effort is enormous.Testing is enormous in any case: =
either the=20
  vendor does it beforeupdate/release or *every* conscientious =
customer does=20
  it beforeupdating production machines.  Naturally, the =
selfish=20
  vendors preferto offload the testing to =
customers.> =20
  Updates are very easy in comparison.  They are easy to apply and =
take=20
  very little time. Only If they don't break
anything.  =
Since they=20
  are not thoroughlytested by their creators, customers are forced =
to either=20
  test them ina lab environent or to roll the dice by deploying an =
untested=20
  fix to aproduction environment.My selfish
viewpoint: It is =

  (e.g.)  Microsoft's product, let them testtest
it.  They =
charge=20
  me enough for software without requiring me totest updates that do =
no more=20
  than fix holes in the products they soldme in the first place.=20
 


------=_NextPart_000_00C2_01C2C87A.BD2FB1A0--

--- BBBS/NT v4.01 Flag-4