[haiku-development] Re: Proposal from Begeistert: Getting Ready for the First Beta

  • From: Niels Sascha Reedijk <niels.reedijk@xxxxxxxxx>
  • To: "haiku-development@xxxxxxxxxxxxx" <haiku-development@xxxxxxxxxxxxx>
  • Date: Tue, 17 Sep 2013 01:10:14 +0200

Hi all,

This message purely reflects my own opinion and does not imply to be the
consensus from BeGeistert in any way.

In my view the opinion to move to a swift release point is based around the
argument that if we critically examine who we are releasing for, then the
conclusion would be that it would be for ourselves. End users? There is a
small group of people that are adventurous and that run the nightlies in
the emulator, they might benefit from a clear milestone. However, the
potential user base that was there 3-5 years ago has probably moved on to
rooting their phone or loading it with custom ROMS.

Commercial software companies and developers? The past five years we have
seen a dramatic shift in development models, many showing that the popular
tools are going to a subscription business model, and others (often
independent developers) fighting over the attention span and dollars of
users in a closed app store. For neither of those Haiku will be a target
platform, and very likely will not be interesting in the near future.

In this context it makes sense to me to celebrate the milestone of package
management by officially acknowledging that with a release. Why? To me it
is the completion of the development platform of Haiku: we now are in the
position to tell app developers to start targeting the system. They are now
able to start developing for Haiku, knowing that if they use the R5 API
together with some non-experimental improvements, and if they package their
software the right way, it will continue to work on whatever final release
we will go to.

Added to that is my observation that for the past year(s) Haiku development
has been stagnant due to the fact that in my impression we were all waiting
for the few that were working on finishing package management. I think now
is the time to wake people up from sleep mode, and get them on their feet
to start testing Haiku on their hardware, and to report and close tickets
on dev.haiku-os.org. My guess is that if we would hold the train for some
desired features that are developed by one or two developers with no
definitive timeframe, we would have to wait another few years to be ready.

Three notes about topics that come up:

* Timing. Jessica, Premislaus and Stephan seem to mention that the
timeframe is too short to deliver the stability we would require. The
timeframe of 2-3 months was actually based on the opinion of some at
BeGeistert that argued that people need to be kicked into release mode, and
that 2 months seems to be the time point where people would be. Any longer
will probably only get people active the last 2 months.

My take: I think that appointing a release manager with veto power will
provide a safeguard against releasing a buggy version too soon.

* Feature X or Y are not ready. Matt mentioned several blockers, Prenislaus
mentioned IMAP and WebPositive.

My take: given the view that the target are the current community, and
currently these issues do not keep us from using it, why would we hold the
release?

* The naming. Should we call it beta with feature X and Y missing?
Shouldn't we first do another alpha?

My take: While this would not be a deal breaker for me, I do feel strongly
about the fact that package management is such a milestone, that in terms
of the message out to the world, signing it off with a different name would
be beneficial. A name like 'developer preview' or 'developer release' (I
thought Be used the latter) would work equally well.

I hope this message makes clear the motivation I have for standing behind
this proposal.

Kind regards,

Niels

Other related posts: