TMG-L Archives

Archiver > TMG > 2003-09 > 1062463728


From: Dennis Lee Bieber <>
Subject: Re: [TMG] What's going on with this Grey Out Business
Date: Mon, 01 Sep 2003 17:48:48 -0700
In-Reply-To: <9EACB07EDEF67647B2261D7750EC370A022B0DCD@exchange6.cgcent.miami.edu>


On or about 17:06 09/01/03 a carrier pigeon from Sutton, Scot G. delivered:
>reproducible by apparently many users (myself included). If companies who
>produce software for corporate clients

There is a subtle difference between having the problem occur for
many users, and being able to reproduce the problem on demand so it may be
profiled by test suites and instrumented tracking builds.

I've also seen many situations (under VMS, using FORTRAN 77) where
the mere act of building a debug version of the code results in the error
going away... until one rebuilds the distribution version. This is usually
a sure indicator that one has not declared an asynchronous shared memory
update -- the distribution version "optimizes" the code, and assumes that
the memory is only used locally and can be stored in a register instead of
memory... so the routine never rereads the memory to pick up changes that
had been made elsewhere.

>At this time WG would appear to be the best software of its kind on the
>market (if it ever gets fully operational) but that may soon be challenged
>by some of the up and coming programs. If WG can't fix the bugs in a
>timely manner, they will start losing clients.

If it hadn't been for the demands of TMG4 users to have /anything/
TMG5 wouldn't be available to be receiving these reports. Anyone using TMG
5 as it currently exists does so with the knowledge that they are acting as
informal testers.

>
>Disclaimer: Both myself and my wife have been compuer programmers for many
>years. If we, as employees took the attitude that we did not wish to find
>and fix such a bug, we would soon be looking for work elsewhere.

Would you agree to expending time solely on a bug reported to you
by someone using a pre-release version of software, and who can not supply
a reproducible sequence of steps by which a development team may create the
bug as needed with a debug build of the software? Especially if you still
have other features that need to be coded to complete the promised
capabilities.

Now, earlier in this batch of messages I see someone has recorded
a sequence of operations which produced the grey-out, and that they were
able to duplicate the grey-out with the same sequence. That information
will no doubt put the grey-out problem on a higher priority for resolving.

--
> ========================================= <
> | Dennis Lee Bieber <
> PGP keys available | RSA keys preferred <
> ========================================= <
> Home Page: http://www.dm.net/~wulfraed/ <



This thread: