1995-10-14 - Re: Netscape question transformed

Header Data

From: Alan Olsen <alano@teleport.com>
To: cypherpunks@toad.com
Message Hash: da5d79e6e09b04ef3f1357c06af56c8172e6598c7ce1565517442a0987b490f6
Message ID: <199510142325.QAA10194@desiree.teleport.com>
Reply To: N/A
UTC Datetime: 1995-10-14 23:25:18 UTC
Raw Date: Sat, 14 Oct 95 16:25:18 PDT

Raw message

From: Alan Olsen <alano@teleport.com>
Date: Sat, 14 Oct 95 16:25:18 PDT
To: cypherpunks@toad.com
Subject: Re: Netscape question transformed
Message-ID: <199510142325.QAA10194@desiree.teleport.com>
MIME-Version: 1.0
Content-Type: text/plain


At 11:12 PM 10/13/95 -0700, jsw wrote:
>zoetrope@infinet.com wrote:
>>         Why would a compony like Netscape jeapordise themselves by even
>> letting                                               out a beta version so
>> hackers have time to find bugs before the full version is released. Don't
>> take me wrong I have only been on the net fer a little over a year and Love
>> Netscape.  And If I knew what I was Doing I would love to help them.....
>
>  The reason we do it is that without wide exposure outside the
>company, it is virtually impossible to get complete testing coverage.
>Given the incredible variety of machines, OS's, peripherals, etc.
>that are available in the real world, there is no way we could
>re-create every possible configuration in the lab.

I think that the main reason people have problems with the testing offer on
the beta is the lack of responses to bugs in the past.  I have seen many
complaints from people on the net complaining that bugs they had reported in
.90 had continued on into new and continued versions.  That if they recieved
a message back from Netscape it was just an infobot, and a rather hostilly
worded one at that.  (The infobot has been lightened up a great deal.)
Netscape has never had the staff to properly respond to the mail volume they
have recieved.  They have been trying to improve this but it is difficult to
man such a project with people who are able to sort through the piles of
dreck to find those people reporting serious problems.  (Also, the bug
rewards are only for "serious bugs".  Who defines what is serious and what
is not.  I know of a couple that have been hanging around for a couple of
versions now.  Maybe I should test if they got fixed with 2.0b1.  Probibly
not.  They only hit a handful of people.)

Netscape is still a pretty new company.  Some seem to forget that .90 was
only released about a year ago.  How long does it take to get a proper QC
department in place?  (After the shakeup in February(?) when the new
President was hired, I would have thought it would have been done then.  Now
I am not so sure...)  I know of at least one product that went out the door
in recient months that was not properly tested.  (The stack and dialer used
with Netscape PE comes to mind.)  I realize that sales sometimes forces
things to be released before its time, but there has to be some basic
controls for quality control.  (I will not mention some of the bigger gaffes
with 1.0 and 1.1 since not all of those were publicized.  A few simple
things could have caught them before they went out the door.)

I want to see Netscape succeed, but when you hear about things like product
being released on the net without warning those people responsible for
supporting it, bug reports being ignored so a ship date can be made, and the
like, it makes you really wonder.  I am finding problems with 2.0b1 that
could have been found just by using the product for a couple of days!  Most
of the problems I have seen with Netscape could be solved with better
communications between departments within the company and some strong
testing procedures.

>  We also can't visit every web page, or think of every combination
>of features that clever people in the real world come up with every
>day.  

Somethings are real simple.  Things like giving it to a handful of people
and letting them use it for a couple of days before you release it to the
rest of the world.  90% of the bugs will be pointed out through having
someone beat on it as they would any other peice of software.  And those
people should be someone other than the programmers.  (Sometimes you get too
close to the code to really stress it properly.)

I agree that you need to have a beta program.  Not everything can be tested.
But QC is important.  This is not Uncle Jim's pet project anymore...

All of this has been brought up on the net before.  Sometimes it seems like
the responses are more defensive than trying to find out the real problems
and fix them.  So far, you have been one of the few who has pushed for
getting some of these things fixed in a timely manner.  (I can name a few
others as well.  They seem to get burned out after a while though...)

I think my biggest questions right now are: "Is there a formal testing
procedure for product that is about to go out the door?  Is there a real QC
department and do they have the power to stop releases if significant
problems are found?"
|             Visualize whirled keys              | alano@teleport.com   |
|"It's only half a keyserver. I had to split the  | Disclaimer:          |
|other half with the government man." - Black Art | Ignore the man       |
|   -- PGP 2.6.2 key available on request --      |  behind the keyboard.|
|         http://www.teleport.com/~alano          |       <fnord>        |






Thread