httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Arved Sandstrom <>
Subject Re: On 2.0 again.
Date Mon, 19 Feb 2001 23:40:15 GMT
At 05:30 PM 2/19/01 -0800, wrote:
Before that, Bill Stoddard held forth:
>> Humm, I believe I implicitly assume a release status of "beta" is
basically a
>> statement regarding "quality" of the server ("Quality" in the sense of "Zen
>> and the Art of Motorcycle Maintenance").  Is this not the model we are
>> following? My goal was to get a threaded server running to facilitate
>> bugs out of that MPM; improving the quality of a major feature of Apache
>> If "beta" status is not a statement of quality, then I don't understand
why we
>> go to the trouble of making the designation in the first place. Perhaps I
>> should write up a neat little random naming facility that makes up status
>> names. Next release can be alpha 1000, then maybe beta 75, then golden
>> followed by alpha centauri :-)
>I am having a hard time understanding why we are requiring our first beta
>to be seg fault free as well.  The server works.  It has been running on
> for at least four or five days.  This is not GA code.  We are
>stable, and things seem to work.  A beta cycle means just that.  We
>believe this is better than alpha code, but not quite completely finished,
>use at your own risk.
>If we wait until the code has zero problems, then what is the benchmark
>for GA code?

Beta and alpha designations are essentially neutral in terms of what they 
explicitly say about software "bugginess". Although both presuppose that 
most of the real testing at that stage is complete.

A real beta is a secondary software test involving selected participants 
before the product is finally released. We expect targeted feedback from the 
beta testers, which is why they are selected in the first place. The testers 
come from outside the developing organisation.

Alpha testing is done by users internal to the developing organisation; the 
software is typically (although not necessarily) at an earlier stage of 
development, and may lack some features.

_Neither_ test is meant to find bugs: unit testing, integration testing, 
system testing, and performance testing should already have done that. 
Strictly speaking neither test is a useability test, either; for that you 
want trained useability testers. So what are they? Real-world 
exposure...introducing that extra random element that you couldn't capture 
in controlled testing, and seeing what it does. I wouldn't be wanting a beta 
tester to report back that the program crashed in flames...that ain't the

I think Bill is more right than you are, Ryan - a beta doesn't mean "use at 
your own risk". It _is_ a statement of quality, more like "hey, here's a 
first look at our practically finished product - let us know what you think!" 

Observation: not very many operations or people use these terms correctly. 
In fact, most software developers I see don't have the first clue what these 
terms really mean. For them a beta usually means "let's perpetrate our 
system test on real users".

Not being condescending in the slightest. I take an interest in formal 
testing and formal QA, and I'm slightly anal over correct terminology. I'm 
not involved with httpd, but since the terms were being bandied about, I 
thought I'd stick my oar in.

Arved Sandstrom

Fairly Senior Software Type
e-plicity (
Wireless * B2B * J2EE * XML --- Halifax, Nova Scotia

View raw message