Re: 1.3a1 Status

["Roy T. Fielding" <fi...@kiwi.ics.uci.edu>]

>I feel that we've reached a point where we can comfortably roll a
>1.3a1. There are a few bugs, but Apache works fairly well on both Unix
>and Windows.

How fast does it run?  I'm curious (but not curious enough to install NT).

>Here's my current plan:
>
>* In a few hours (about 5:00 PM PDT), I'll tag 1.3a1, and generate .zip,
>  .tar.gz and .tar.Z files for review by the Group, unless someone
>  objects in the interim.
>
>* Assuming these pass review, tomorrow around noon (PDT), I'll move them
>  to the live site, and update the various web pages. I'll leave it to
>  others to make announcements to mailing lists/newsgroups, if they feel
>  that neccessary.
>
>Sound good to anyone?

Ummm, not really.  Since when did we start releasing alphas?  For that
matter, when did we decide that alpha numbering was a good idea?
I don't even like beta numbering, so you can imagine how I feel about
yet another ambiguous numbering scheme.

Why don't we just number our releases like sensible folk and attach
a relative-quality attribute to the name, e.g.

    apache-1.3.0-alpha.tar.gz 
    apache-1.3.1-alpha.tar.gz 
    apache-1.3.2-beta.tar.gz 
    apache-1.3.3-beta.tar.gz 
    apache-1.3.4.tar.gz 

BTW, you forgot to change APACHE_RELEASE for 1.3a2-dev (whatever).

....Roy

Re: 1.3a1 Status

[Dean Gaudet <dg...@arctic.org>]

On Tue, 22 Jul 1997, Roy T. Fielding wrote:

>     apache-1.3.0-alpha.tar.gz 
>     apache-1.3.1-alpha.tar.gz 
>     apache-1.3.2-beta.tar.gz 
>     apache-1.3.3-beta.tar.gz 
>     apache-1.3.4.tar.gz 

FWIW, and I don't want to stir up this debate again, I prefer numbered
systems like this.  Being a linux weenie I like linux' numbering scheme... 
second digit odd means it's not stable.  Yes I understand the
deficiencies. 

As far as releasing an alpha, we did decide to do it a while ago.  The
essential difference is that we're not in feature freeze yet, we want
feedback on new features (such as NT). 

Dean

Re: 1.3a1 Status

[Alexei Kosut <ak...@organic.com>]

On Tue, 22 Jul 1997, Roy T. Fielding wrote:

[Forgot to respond to this part]

> BTW, you forgot to change APACHE_RELEASE for 1.3a2-dev (whatever).

If you want things like this done, you either need to do them yourself,
or add them to http://dev.apache.org/how-to-release. No one can rememeber
these things otherwise.

-- Alexei Kosut <ak...@organic.com>

Re: 1.3a1 Status

[Brian Behlendorf <br...@organic.com>]

At 10:27 AM 7/23/97 -0700, Roy T. Fielding wrote:
>>Um... Roy, we decided to release an alpha of 1.3 at least ten days ago,
>>and we've been referring to it as 1.3a1 for at least a week.
>
>Sorry, I missed that -- I've just been skimming the NT mail.  In any
>case, it doesn't make any sense to announce an alpha release, since
>alpha releases are not intended for use by customers.

As a freeware project I think we have a fair bit more liberty towards
"showing our warts" than companies who have customers (which implies
payment).  In the announcement message (which I'll draft when I get back
from lunch), we'll make it clear that this type of release is unprecedented
for this group, but we wanted to make it so that people can see one
direction we're taking.  It'll be clear that this release is not for public
consumption, that folks should not put their mission-critical sites on it,
but they should check it out and give us some feedback.  Since it's not a
binary release, we don't have to worry as much about a huge wave of bug
reports.  We will not be recommending people upgrade, or telling folks we
won't support 1.2.  I hope the Subject: line of the announcement is not
"Apache 1.3a1 released!".

I don't think we'll see more than two alphas.

>NCSA httpd never used beta numbering for public releases,
>and NO software is *publically released* with an alpha designation.

Not to distract too much, but I definitely remember a 1.3b leading to a
1.3... 

>Alpha is commonly used as a derogatory term for software that should not
>have been released at all.

Perhaps there's a middle point; perhaps instead of announcing on the
apache-announce list, we play it more subtle.  Post it to the site, post a
message about this to a couple NT developers groups, etc.  Personally I'd
like to send it to the whole list, with heavy police caution tape all over it.

	Brian


--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=--
"Why not?" - TL           brian@organic.com - hyperreal.org - apache.org

Re: 1.3a1 Status

[Alexei Kosut <ak...@organic.com>]

On Tue, 22 Jul 1997, Roy T. Fielding wrote:

> >I feel that we've reached a point where we can comfortably roll a
> >1.3a1. There are a few bugs, but Apache works fairly well on both Unix
> >and Windows.
> 
> How fast does it run?  I'm curious (but not curious enough to install NT).

Wish I knew... point me at a benchmark program that works, and I'll tell
you.

[...]

> >Sound good to anyone?
> 
> Ummm, not really.  Since when did we start releasing alphas?  For that
> matter, when did we decide that alpha numbering was a good idea?
> I don't even like beta numbering, so you can imagine how I feel about
> yet another ambiguous numbering scheme.

Um... Roy, we decided to release an alpha of 1.3 at least ten days ago,
and we've been referring to it as 1.3a1 for at least a week.

> Why don't we just number our releases like sensible folk and attach
> a relative-quality attribute to the name, e.g.

Because that's not how "sensible folk" have released software for nearly
twenty years now? What we're doing is, AFAIK, what software engineering
people have done forever. Your plan would be really confusing to people
(myself included).

-- Alexei Kosut <ak...@organic.com>