Hi,

I am totally with you. We should automate the release-related tooling,
the creation of the related notes as well as the download pages. 

However, we should keep in mind, that we had some Antorra-based website
rework. I do not remember the current status but I guess it is also
.adoc based - so I guess, it would be no show stopper.

Might also a good starting point for some contributors out where :-)

Gruss
Richard

Am Mittwoch, den 24.03.2021, 15:14 -0700 schrieb David Blevins:
> > On Mar 24, 2021, at 12:40 AM, Richard Zowalla <[email protected]>
> > wrote:
> > 
> > I have merged the "manual" changes to add the missing links to the
> > .asc
> > files of the release download page.
> > 
> > In this process, I discovered, that we are not using the Downloads
> > generator class [2] anymore. My questions is:
> > 
> > Did we switch on purpose from Maven repository links (for the
> > binaries)
> > towards mirror links? (just want to catch up on this)
> 
> We definitely should reinstate the automation, but we probably want
> to redo the Downloads generator class.
> 
> Someday, hopefully in the near future, we'll be some form of Jakarta
> EE certified.  When that happens we'll need to create a "public test
> results page" for each release.  It needs to contain the following
> information[1]:
> 
> - Product Name, Version and download URL
> - Specification Name, Version and download URL
> - TCK Version, digital SHA-256 fingerprint and download URL
> - Java runtime used to run the implementation
> - Summary of the information for the certification environment,
> operating system, cloud, …
> - List of tests passed (can be a link to another page, say the xml
> output from running the TCK)
> 
> I drafted those requirements that were eventually approved and
> adopted.  What was in my head is that smart implementations would
> simply use their actual download page for these details.  Our current
> table format isn't that great in the first place (kind of noisy and
> cluttered), so not really a good starting point for adding more.
> 
> In the distant past we used to create a page for each release, even
> including generated release notes from JIRA.
> 
>  - http://tomee.apache.org/tomee-1.0.0-beta-1.html
>  - http://tomee.apache.org/tomee-1.5.0.html
>  - http://tomee.apache.org/tomee-1.6.0.1.html
>  - http://tomee.apache.org/tomee-1.7.0.html
> 
>  - http://tomee.apache.org/tomee-1.5.0-release-notes.html
>  - http://tomee.apache.org/tomee-1.5.1-release-notes.html
>  - http://tomee.apache.org/tomee-1.7.0-release-notes.html
> 
> The code that generated the release notes (and most the release tasks
> even including vote threads) is here:
> 
>  - 
> http://svn.apache.org/repos/asf/tomee/sandbox/release-tools/src/main/java/org/apache/openejb/tools/release/cmd/
> 
> This just decayed over time.  I'd be blissfully happy with a single
> generated page for each release that contained just the download
> links for now.  In the future we can add certification information
> below the download links.  In the future future, we can add generated
> release notes in list format below that.
> 
> There's just so much automation we could be doing around releases,
> especially now that we could potentially setup Jenkins to do the
> work.  We could have parameterized jobs to do nearly everything that
> was going on in the `asf/tomee/sandbox/release-tools/` repo.
> 
> Tasks like adding the new release to the mirror and removing the old
> one could be automated.  As could generating a release page and
> pushing it to the website, while also updating the page for the just-
> removed-from-the-mirror release to now point at archive.apache.org.
> 
> At tomee.apache.org/download we could either point to the latest
> release page and have no direct zip, tar links or have some sort of
> short version of the download links.  I'd probably be in favor of
> just pointing so it's easier to maintain.
> 
> On the release notes, what killed that is all the JIRA APIs changed
> from RPC, to some xml format to now REST.  The library I wrote in the
> 2006-range (swizzle-jira) eventually got too outdated and stopped
> working around 2012.  There are now Atlassian maintained java clients
> for JIRA as far as I understand, so we'd could potentially make
> something new with that.
> 
> Of course we can just fix the current Downloads.java for now, but if
> you get inspired it'd be a thing of beauty to see the project become
> what it once was :)
> 
> That said, we have a brief window from now till end of April before
> Jakarta EE 9.1 is scheduled to be released.  I don't know if we can
> get completely certified in that time, but we're down to 400-500
> failing tests on EE 9.1 and at least a hundred or two of them are
> tied to changes we'd need to make in the TCK which are only possible
> between now and end of April.  After that we'd need to file TCK
> challenges to have tests removed, but can't actually fix them.  I'll
> be doing what I can in the time I can get to push on that and help is
> always welcome!
> 
> So many things to do, so little time! :)
> 
> 
> -David
> 
> 
> [1] https://jakarta.ee/committees/specification/tckprocess/
-- 
Richard Zowalla, M.Sc.
Research Associate, PhD Student | Medical Informatics

Hochschule Heilbronn – University of Applied Sciences
Max-Planck-Str. 39 
D-74081 Heilbronn 
phone: +49 7131 504 6791
mail: [email protected]
web: https://www.mi.hs-heilbronn.de/ 

Attachment: smime.p7s
Description: S/MIME cryptographic signature

Reply via email to