Skip to Content.
Sympa Menu

sm-discuss - Re: [SM-Discuss] Let's make Source Mage not suck!

sm-discuss AT lists.ibiblio.org

Subject: Public SourceMage Discussion List

List archive

Chronological Thread  
  • From: Jeremy Blosser <jblosser-smgl AT firinn.org>
  • To: sm-discuss AT lists.ibiblio.org
  • Subject: Re: [SM-Discuss] Let's make Source Mage not suck!
  • Date: Sun, 7 Jan 2007 14:24:41 -0600

On Jan 07, Andrew [afrayedknot AT thefrayedknot.armory.com] wrote:
> On Sun, Jan 07, 2007 at 10:52:49AM +0500, Alexander Tsamutali wrote:
> > Users want to install operating system, install stable software and
> > then they want to install only critical or security bugfixes (support).
>
> What users? How many? Where are they? How come no one else has said
> anything?

It's worth noting that this thread is not yet a week old and it's still the
holiday season. Given the low level of activity recently it's entirely
likely that not everyone has even seen this discussion yet. I'm *not*
saying that's a reason to not move forward on these ideas but I don't know
that an argument for consensus from people's silence is completely
compelling right now.

> We have limited resources, you must accept we cannot please everyone,
> given that others have shown support, or at least given silent approval,
> that indicates that 2 weeks will fit some of our requirements and please
> a fair number of people. If it didnt, then they wouldn't respond.
>
> > After some time passes they want to upgrade operating system, and i
> > don't think they want to upgrade every 2 weeks, so support time should
> > be longer (at least 6 month, better more than a year).
>
> First of all, who are these users? We need evidence.

We hear from a decent number of users who do the 6 month / 1 year update
thing, though we've never really accepted supporting that timeframe.
According to

http://ledger.sourcemage.org/users/statistic_live.php?select=openssh

6 of 41 machines (14%) are running openssh 4.3p1, which hit the grimoire
2006-02-01 and is older than what's even in the current stable. This isn't
a lot of data (it would be nice if ledger could show the grimoire those
boxes are on; does anyone know a way?) but it's what we probably have
available.

While people running stable in production are going to be the most
critical cases (as you noted), we hear it from general workstation users as
well.

> I think we can say with confidence that many users are not pleased by
> a stale 6-12 month old grimoire.
>
> I'll go one further and say that I think users will like short quick
> updates once every two weeks rather than massive yearly ones. I dont
> update many of my machines currently because the stable updates are
> always so substantial.

There's 2 majorly conflicting use cases here: there's people that want all
the latest stuff and don't mind small, frequent updates, and there's people
that cannot update very often for whatever reason and then have issues if
the update is too substantial when they finally get a chance, or worse,
they've missed some updates and we've pulled some of the transitional code
and they can't do a clean update. There's going to be a range of opinions
on where on that continuum we should put "stable"; some people think they
deserve the bleeding edge even if running "stable", and some people won't
even believe something that's been out only 2 weeks can fit a definition of
"stable". Security updates aren't included in that discussion; most
people accept those should get applied ASAP and if they disagree we dont'
really care.

However like you noted the bottom line right now has to be that:

> We dont have a lot of resources, you need to keep that in
> mind, no implementable plan will please everyone, I would go further and
> say that no plan, implementable or not can please everyone.

This is definitely true and I think in the end we have to take whatever
approach works for our *developers* first, because as few as we are
anything that won't work for them will mean we don't do any updating at
all, and that pleases no one. But I do think that whatever we decide on
needs to be communicated to our users so they can figure out how to manage
themselves.

> Also, theres a subtlety here that you have overlooked. We dont support
> intra-stable updates. However we provide archival versions of stable.
> So someone could upgrade serially, one release at a time during one
> "upgrade session".

Where do we provide archival versions of stable? This is definitely a good
way to address this issue, but to my knowledge we don't provide old
grimoire tarballs right now. This is bug #11809; there it's primarily
meant to address the "old versions of spells" issue, but we might extend it
for this by making sorcery upgrade grimoire-version-aware warning people
trying to skip too many versions that they should do it serially.

A very related issue is that we don't currently have a solid definition of
how long transitional code should stick around. It used to be two weeks,
which we had to change when we got off the two week cycle, now it's loosely
seen as 2 stable releases. We need to define that for the benefit of
users. Something like the above recommended upgrade path would make this
easier to define.

> Many years ago, before "devel" was taken offline, we had a two week
> release cycle. Devel became test after two weeks, test became stable
> after two weeks. That was flawed because we didnt formally test test.
> However a precedent exists for a two week cycle.
>
> Lastly, since I've been a part of this distro, we've always recommended
> updating on a regular basis, regardless of the grimoire you chose.

We do recommend it and always have but I think we can do our users the
favor of defining "regular basis" and note risks they'll run if they go
outside of that, especially since our own definitions have been changing as
we try to figure out what works.

Attachment: pgp6jkqmdrthW.pgp
Description: PGP signature




Archive powered by MHonArc 2.6.24.

Top of Page