The maintenance issue is a tricky one with no right answer for all consumers. Some want new features, some want bug fixes for this year’s release, some want bug fixes for last year’s release, some want bug fixes for the release three year’s ago, etc.
Ideally, out aggregation processes and infrastructure should support creating new aggregation streams with different goals and rules, similarly to how we can create new EPP packages simply by having an interested party step forward to own it.
This intersects with the topic of properly supporting check for updates and ensuring that no aggregation participant is injecting their own update site into check for updates list (currently under discussion at the architecture council), otherwise users will accidentally hop off the stream that they started on.
- Konstantin
From: Ed Willink
Sent: Thursday, September 24, 2015 8:13 AM
To: cross-project-issues-dev@xxxxxxxxxxx
Subject: [cross-project-issues-dev] Maintenance builds (was Announcing a oneweek slip in the Mars.1 release (from 9/25 to 10/2))
Hi
That makes sense but shows that we are just shifting the problem.
I see a requirement for
- regular base releases (yearly)
- maintenance releases (four monthly)
- responsive releases (four monthly)
Recognising that maintenance releases were being abused to provide responsive releases is probably good, but waving goodbye to maintenance releases is bad.
IMHO we need all three and so long as we try to make do with two we will be in trouble with some user community. It seems wrong that because some projects have abused the principles of maintenance, users of other projects that have observed maintenance discipline suffer.
Regards
Ed Willink
On 24/09/2015 15:35, Ian Bull wrote:
Ed,
The reason for the change from Mars SR1 to Mars 1 is because this is how we've been doing it for years. Many people (EGit / JGit, Mylyn, CDT -- to name a few) had been putting minor releases in the release train during the SRs. I ran some numbers last year, and > 1000 Installable Units had incremented their minor version number between SR0 and SR2. This means, assuming people are following the version guidelines, that up to 1,000 bundles had already been adding new API between SR0 and SR2.
Changing the name of the train just means we are acknowledging what was already happening.
On Wed, Sep 23, 2015 at 11:21 PM, Ed Willink <ed@xxxxxxxxxxxxx> wrote:
HI
Surely this is the inevitable consequence of Mars.1 rather than Mars SR1?
SR1 required each component to be a safe upgrade so that exact release timing was irrelevant.
Mars.1 is a new release so users must get to see the co-ordinated new release in one go rather than incrementally. If A.1 pulls in B.1, but C uses B, users of C are in a mess until they get C.1.
Regards
Ed Willink
On 24/09/2015 05:26, Gunnar Wagenknecht wrote:
David,
Am 23.09.2015 um 23:37 schrieb David M Williams <david_williams@xxxxxxxxxx>:
If not obvious, this means all participants in "coordinated release train" should not make your releases visible on 9/25, but wait until 10/2 10 AM to make them visible, and announce your official releases.
This seem unnecessarily restrictive. I don't bother with the announcement part. However, I don't recall there is something in the process that requires project to wait publishing the release bits. Not making them visible could have a huge effect on a project's adopter community.
-Gunnar
--
_______________________________________________
cross-project-issues-dev mailing list
cross-project-issues-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://dev.eclipse.org/mailman/listinfo/cross-project-issues-dev
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2015.0.6140 / Virus Database: 4419/10692 - Release Date: 09/24/15
