1.3a1

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

We had discussed releasing a 1.3 public alpha on the 15th, which is
tomorrow. I don't think at the moment that would be appropriate.

Ben is in the middle of reorganizing the NT source and makefiles to make
use of the DLL capabilities I submitted last week, and hasn't
finished. It would be very confusing to release the server at this
point.

Also, there is little or no documentation on using the Windows version of
the server. There's Ambarish's README.NT file (modified slightly by Ben),
which applies to an older version of the code and is less than completely
helpful to the current situation. I plan to write some docs on basic
Apache/Windows compilation/setup/configuration, but can't do so until the
dust from Ben's reoganization settles, and some things are still up in
the air.

So, assuming Ben finishes quickly, how about Friday? That should be
enough time to get everything set (I won't be available most of Thursday,
anyway, and Wednesday is too close). I'll volunteer to package up the
release, if no one else wants to.

Also, there's a question specific to this alpha release: In addition to
releasing the source package, as we always have, we've discussed
releasing a Windows-specific binary package that (unlike the Unix
sources) contains the binary, no source, and DLL versions of the optional
modules. This would be for the vast majority of Windows users who not
only don't want to compile the server, but don't own a compiler, don't
know what one is, and want nothing to do with it.

The question is whether we want to do this for 1.3a1. In other words, do
we want to make this release of Apache available for all Windows users,
or just those with compilers? The rationale for the latter would be to
cut down the testers to those who presumably know what they're doing and
would report bugs in a realistic fashion. It would, however, cut down
substantially on the amount of people who might try out the package.

Any thoughts?

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

Re: 1.3a1

[Ben Laurie <be...@algroup.co.uk>]

Alexei Kosut wrote:

> So, assuming Ben finishes quickly, how about Friday? That should be
> enough time to get everything set (I won't be available most of
> Thursday,
> anyway, and Wednesday is too close). I'll volunteer to package up the
> release, if no one else wants to.

The timing is unfortunate - my brother has just returned from, ahem, a
certain event, and we have much to discuss, so this evening won't see
much done - but probably some. Tomorrow I have guests, so Thursday looks
like being the earliest I am likely to finish. But I'll see what I can
do!

> Also, there's a question specific to this alpha release: In addition
> to
> releasing the source package, as we always have, we've discussed
> releasing a Windows-specific binary package that (unlike the Unix
> sources) contains the binary, no source, and DLL versions of the
> optional
> modules. This would be for the vast majority of Windows users who not
> only don't want to compile the server, but don't own a compiler, don't
>
> know what one is, and want nothing to do with it.
>
> The question is whether we want to do this for 1.3a1. In other words,
> do
> we want to make this release of Apache available for all Windows
> users,
> or just those with compilers? The rationale for the latter would be to
>
> cut down the testers to those who presumably know what they're doing
> and
> would report bugs in a realistic fashion. It would, however, cut down
> substantially on the amount of people who might try out the package.

1.3a1 should be source only. Once the source only releases stop
producing substantial feedback, we go for binaries.

Cheers,

Ben.

--
Ben Laurie [Normal service will be resumed as soon as possible]

Re: 1.3a1

[Alexei Kosut <ak...@nueva.pvt.k12.ca.us>]

On Mon, 14 Jul 1997, Rasmus Lerdorf wrote:

> One thing we might need to establish is how many people within the group
> here, have actually compiled and tried the Win32 version.  Ben, Alexei,
> Ambarish (obviously) and myself are the only ones I know of.  I think that
> if within this group we can only pull together 4 people to test the source
> version, we are going to have a really tough time getting a lot of outside
> testers.  And if we do release a source version, I have a feeling that

Except that the Apache Group is, on the whole, a Unix-based
development team. No one joined the group (except recently) with the
expectation of working on Windows; I imagine there are large hordes of
Windows developers who would never before have even thought of working
on Apache who now will.

> within hours there will be all sorts of binary versions floating about
> anyway.  We may as well keep some control over how the thing is compiled
> and release a binary version so we have some hope of determining whether a
> reported problem is a genuine bug or somebody who didn't know how to use
> their compiler.

That is a valid point, though.

-- 
________________________________________________________________________
Alexei Kosut <ak...@nueva.pvt.k12.ca.us>      The Apache HTTP Server
URL: http://www.nueva.pvt.k12.ca.us/~akosut/   http://www.apache.org/

Re: 1.3a1

[Paul Sutton <pa...@ukweb.com>]

On Mon, 14 Jul 1997, Rasmus Lerdorf wrote:
> One thing we might need to establish is how many people within the group
> here, have actually compiled and tried the Win32 version.  Ben, Alexei,
> Ambarish (obviously) and myself are the only ones I know of.  I think that

Ahem, I've been using the NT version since Ambarish posted it.  Even wrote
the (minimal) readdir() replacement that isn't GPL copyleft. But I'm
really not sure about the mixed multi-threaded/multi-process model. What
happens to threads currently handling a long slow tranfer when the child
dies?

//pcs

Re: 1.3a1

[Rasmus Lerdorf <ra...@lerdorf.on.ca>]

> The question is whether we want to do this for 1.3a1. In other words, do
> we want to make this release of Apache available for all Windows users,
> or just those with compilers? The rationale for the latter would be to
> cut down the testers to those who presumably know what they're doing and
> would report bugs in a realistic fashion. It would, however, cut down
> substantially on the amount of people who might try out the package.
> 
> Any thoughts?

One thing we might need to establish is how many people within the group
here, have actually compiled and tried the Win32 version.  Ben, Alexei,
Ambarish (obviously) and myself are the only ones I know of.  I think that
if within this group we can only pull together 4 people to test the source
version, we are going to have a really tough time getting a lot of outside
testers.  And if we do release a source version, I have a feeling that
within hours there will be all sorts of binary versions floating about
anyway.  We may as well keep some control over how the thing is compiled
and release a binary version so we have some hope of determining whether a
reported problem is a genuine bug or somebody who didn't know how to use
their compiler.

-Rasmus

Re: 1.3a1

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

On Mon, 14 Jul 1997, Marc Slemko wrote:

> Sounds good to me.  
> 
> How many Windows binaries do we need?  Will it run on 95 with one binary
> for all Intel NT and 95 platforms?  3.51 and 4.0? 

One binary should run on all Win32 OSes. This includes Windows 95,
Windows NT 3.51 and Windows NT 4.0. I've minimally tried all three,
although most of my use has been on NT Server 4.0.

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

Re: 1.3a1

[Marc Slemko <ma...@worldgate.com>]

Sounds good to me.  

How many Windows binaries do we need?  Will it run on 95 with one binary
for all Intel NT and 95 platforms?  3.51 and 4.0? 

How about Alpha boxes?

On Mon, 14 Jul 1997, Alexei Kosut wrote:

> So the plan looks something like this:
> 
> 1. Sometime this week (maybe), build a 1.3a1 source package. Put a
>    message on www.apache.org announcing it, being very cautious about the
>    wording. ("This is still prerelease software, but anyone wishing to
>    try out Apache with Windows or the new Unix features may use it at
>    their own risk.") Don't announce anywhere else.
> 
> 2. If it seems to work for people, maybe in another few weeks we'll
>    release an 1.3a2, with an Windows-binary version as well. Announce this to
>    all the normal places. This would be a "not-done" release, but with a
>    Windows port we feel works pretty well.
> 
> 3. Soon after that, declare a feature freeze and release 1.3b1. Normal
>    rules apply, plus Windows binary for each beta release.
> 
> -- Alexei Kosut <ak...@organic.com>
> 

Re: 1.3a1

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

On Mon, 14 Jul 1997, Marc Slemko wrote:

> On Mon, 14 Jul 1997, Brian Behlendorf wrote:
> 
> > If/when we announce it to apache-announce we need to be very careful about
> > cautioning folks on its use.  We should also announce it on various
> > NT-programming specific newsgroups.
> 
> My suggestions is don't announce a1 much at all (you have to announce it
> somehow, but not widely), then wait a few days or a week to see if there
> are any huge problems.  If not, announce more widely.

So the plan looks something like this:

1. Sometime this week (maybe), build a 1.3a1 source package. Put a
   message on www.apache.org announcing it, being very cautious about the
   wording. ("This is still prerelease software, but anyone wishing to
   try out Apache with Windows or the new Unix features may use it at
   their own risk.") Don't announce anywhere else.

2. If it seems to work for people, maybe in another few weeks we'll
   release an 1.3a2, with an Windows-binary version as well. Announce this to
   all the normal places. This would be a "not-done" release, but with a
   Windows port we feel works pretty well.

3. Soon after that, declare a feature freeze and release 1.3b1. Normal
   rules apply, plus Windows binary for each beta release.

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

Re: 1.3a1

[Marc Slemko <ma...@worldgate.com>]

On Mon, 14 Jul 1997, Brian Behlendorf wrote:

> If/when we announce it to apache-announce we need to be very careful about
> cautioning folks on its use.  We should also announce it on various
> NT-programming specific newsgroups.

My suggestions is don't announce a1 much at all (you have to announce it
somehow, but not widely), then wait a few days or a week to see if there
are any huge problems.  If not, announce more widely.

But I don't really care either way...

Re: 1.3a1

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

At 05:06 PM 7/14/97 -0700, Alexei Kosut wrote:
>So, assuming Ben finishes quickly, how about Friday? That should be
>enough time to get everything set.

+1, though I think we should do the code reformatting before then too. (so
I guess +1 on that as well).

>Do
>we want to make this release of Apache available for all Windows users,
>or just those with compilers? The rationale for the latter would be to
>cut down the testers to those who presumably know what they're doing and
>would report bugs in a realistic fashion. It would, however, cut down
>substantially on the amount of people who might try out the package.

Since it's *alpha*, and since we really want to solicit bug reports form
people likely to be able to also solicit patches, I would +1 a source-only
release, at least for a1.  If we get enough complaints maybe we can
consider it.  

If/when we announce it to apache-announce we need to be very careful about
cautioning folks on its use.  We should also announce it on various
NT-programming specific newsgroups.

	Brian


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