Hi Yuki,
On Wed, 9 May 2001 07:03:57 +0900GMT (09/05/2001, 06:03 +0800GMT),
Yuki Taga wrote:
YT>>> From 1.51 to 1.52 is a **minor** version change.
T>> I think the Connection Center is a major change.
YT> If so, then it should have an appropriate version number.
ACK.
YT>>> For most users, there is simply no justification in downloading it
YT>>> and installing it -- unless you want to be constantly on the
YT>>> buggy, cutting edge of development.
T>> This is generally true. But how about new users who download for the
T>> first time from the official page and get a buggy "new features"
T>> version?
YT> RIT should be *especially* careful about this,
ACK - and I was pretty disappointed that they posted the buggy beta
version on the web site, just renamed it 1.52, and keep sending the
subsequent beta versions as release veresions with a letter added to
the web site. Bad style; quite damaging to their reputation, rather.
YT>>> Software construction is a trial and error process.
T>> If that were true, software engineering would be in deep trouble.
YT> As a programmer, I can tell you it is indeed a trial and error process.
Well, as an MSc student in Software Engineering, I hope we are talking
about different degrees of "trial and error in a released program"
here. I agree that a program is never quite finished, but that should
relate to features rather than bugs. You alpha- and beta-test a
program before you release it; the released version should be
virtually bugfree. There will usually still be some minor bugs, but if
you use a sound engineering approach - whichever life cycle you want
to use, there is always the testing stage before implementation - and
make heavy use of the 'reuse' concept (which admittedly is not always
possible to a large degree), the bugs should be so few that a user
would find them only occasionally. If the bugs are found quickly, then
you have to revise your testing strategy!
What we dealt with here, namely the v1.52 "release" series, contained
a number of known major bugs, and that is just unexcusable. A bug in a
release version is a programming mistake that wasn't known before the
release; I hope you agree at least on this one. To sell a version that
has known major bugs is beyond my comprehension.
Example: I hear the bug that people could not choose their DUN
connection in the Connection Center (greyed out under certain
circumstances) has been fixed in v1.52f. I don't know whether it is
really fixed, because the beta testers have no beta to test, and
newbies will find out now. The connection is a core part of an email
client, and this bug was known well before 1.52 was uploaded. What is
your excuse for Rit?
YT> Yes, there is an engineering style approach,
Is there any other professional approach to software development? Oh
yeah, I forgot: trial and error. No ma'am, not for me. ;-)
YT> but this is not bricks and mortar; this is software.
Does this mean it is OK to release a program that has known major
bugs?
YT>>> From 1.51 to 1.52 is not going to improve your sex life. Pretty sure about
YT>>> that.
T>> You are right. How did you know that?
YT> Womens' intuition.
You are a step ahead of me then. I have no access to that resource.
;-)
YT>>> And 1.52x to 1.52y is not likely help much either.
T>> Really? In that case, what is the improvement? ;-)
YT> Does the dishes now. ^_^
I *do* need it! :-)
--
Cheers,
Thomas.
Moderator der deutschen The Bat! Beginner Liste. Anmeldung unter:
[EMAIL PROTECTED]
Message reply created with The Bat! 1.51
under Chinese Windows 98 4.10 Build 1998
on a Pentium II/350 MHz.
--
______________________________________________________
Archives : <http://tbudl.thebat.dutaint.com>
Moderators : <mailto:[EMAIL PROTECTED]>
TBTech List: <mailto:[EMAIL PROTECTED]>
Unsubscribe: <mailto:[EMAIL PROTECTED]>
You are subscribed as : [email protected]
