From mboxrd@z Thu Jan  1 00:00:00 1970
From: ben.hutchings@codethink.co.uk (Ben Hutchings)
Date: Thu, 15 Sep 2016 23:31:34 +0100
Subject: [cip-dev] Introduction
In-Reply-To: <2db652238ec32fb02f30541a75454292@codethink.co.uk>
References: <1473851440.4719.37.camel@codethink.co.uk>
	<005e01d20f1d$64afb270$2e0f1750$@toshiba.co.jp>
	<57DA5C43.5010001@codethink.co.uk>
	<2db652238ec32fb02f30541a75454292@codethink.co.uk>
Message-ID: <1473978694.23987.41.camel@codethink.co.uk>
To: cip-dev@lists.cip-project.org
List-Id: cip-dev.lists.cip-project.org

On Thu, 2016-09-15 at 17:17 +0100, Paul Sherwood wrote:
[...]
  [Greg K-H wrote:]
> "But if we didn't provide an LTS, would companies constantly update 
> their kernels to newer releases to keep up with the security and 
> bugfixes? That goes against everything those managers/PMs have ever been 
> used to in the past, yet it's actually the best thing they could do."

Linux does have a very good record of maintaining application binary
compatibility, which should make it safe to upgrade.  But every new
release from Linus brings some or all of the following problems:

- feature regressions
- performance regressions
- security regressions
- increased resource consumption
- removal of deprecated features
- driver API churn

Most of the regressions get fixed quickly, but some linger long after
support for the previous stable branch has ended.

So I don't think it's generally sensible to use the latest stable update
in production, and that's why I care about longterm branches (and I
don't know why Greg does).

> I've been recommending the constant update route route to customers 
> over the last few years, with some success, but many ecosystem members 
> are extremely uncomfortable with the whole idea of aligning with 
> mainline.  I think this is broadly because as embedded engineers we've 
> learned over many years that it's best to change the platform as little 
> as possible.  I wrote an article trying to challenge this traditional 
> embedded thinking earlier this year [3]
> 
> Would be very interested in others' thoughts on this.

I agree it's possible to do rolling upgrades, but it's very dependent on
the capability and the will (1) to do regular regression testing on the
target systems and (2) to address the regressions that are found without
waiting for them to be fixed upstream.  Where a performance regression
or increased resource consumption stretches an existing system so that
it no longer meets its requirements, this might not be practically
possible.  This also applies where non-upstream code is broken by
upstream API changes, and that issue is not going away soon.

Ben.

-- 
Ben Hutchings
Software Developer, Codethink Ltd.