Should we test running with Java 11? The thought is Java 11=
is out of scope for 5.0. We will create a ticket for it and include =
it in the next release.
Can someone test with Windows?
Samvera / Islandora community =E2=80=93 is anyone there going to b=
e able to test it? It was talked about on the Islandora Tech call, bu=
t unsure of what was talked about. Should touch base with Jared Whiklo and/or Danie=
l Lamb to see what they're thinking, given our timeline.
Andrew to reach out to Trey Pendrago=
n to touch base around Fedora 5 testing. Use case for Valkerie - =
pushing a fully formed data set into Fedora, as described by Michael B. Klein.
fcr:backup/fcr:restore - 4.7.5 data set is not expected to work with=
5.0. This section on backwards compatibility should be revisited.&nb=
sp; We don't promise backwards compatibility for backup/restore.
Danny will look at the Jetty stuff.
WebAC tests - now that we have a robust setup, we should add some tests=
for it to the document. Andrew will clean up the release testing pag=
e a bit and add things like WebAC and External Content to list.
There are good tests for External Content, etc in the API Test Suite - =
that should be added to the Release Test sheet. Danny Bernsteinwill look into doing that.
Peter Eichman will look=
at camel-toolbox and document what it will take to bring that into alignme=
nt with 5.0. He was able to start some minimal testing based on=
his setup with vagrant.
Yinlin Chen can test "PostgreSQL lin=
ux" and will update the Release Test page once he does.
Discussed when to cut RC-2 - more notes on that below.
Unable to locate J=
ira server for this macro. It may be due to Application Link configuration.=
- The end point exists, but not actually created in the repo=
until a memento is created. Danny will create a PR for this.
To defer to 5.1?
Unable to locate J=
ira server for this macro. It may be due to Application Link configuration.=
Unable to locate J=
ira server for this macro. It may be due to Application Link configuration.=
Unable to locate J=
ira server for this macro. It may be due to Application Link configuration.=
Unable to locate J=
ira server for this macro. It may be due to Application Link configuration.=
- this has been there a long, long time.
Unable to locate J=
ira server for this macro. It may be due to Application Link configuration.=
- is this easy to remove? Still valuable to have some=
way for the response to indicate what the user is allowed to do in the Web=
UI. Peter Eichmanh=
as added a comment to the ticket. More notes on this ticket below.
Are we okay with putting all of these off? With the excepti=
on of FCREPO-2938, these tickets will not go into 5.0.
Forking Modeshape: What is our plan for modeshape for LTS? =
=20
Michael B. Klein Modeshape project: it =
does look like someone is accepting pull requests on that project. Un=
clear if they are cutting releases off of master.
What will we do, long-term, if the Modeshape community is being slow in=
responding to PRs?
Aaron Birkland noted that one can=
not work with Modeshape and S3 with out this pat=
ch.
What does it look like if we have a Modeshape fork? =20
We could have a fork that meets our needs and then move on once we remo=
ve Modeshape from the project. It would not be a very long-term commi=
tment to have this fork.
If we fork it and then the project seems to revive itself, we could go =
back to the main Modeshape code base.
Who is in a position to move this forward - under fcrepo4/ in github?&n=
bsp; Andrew Woods has had trouble build=
ing Modeshape in the past - Michael B. Kle=
in could contribute his initial work on this issue, but may not be able=
to maintain it long-term. Aaron B=
irkland has never been able to build it successfully, either. He might =
be able to help work with this fork as well.
Release candidate
APIX Testing - Aaron Birkland=
will try out fedora 5 with APIX and update the test page to let us know re=
sults.
Ecosystem tools:
Java Client - Ben Pennelldid some great work creating an epic of wo=
rk that needs to be done here.
Docker - Yinlin Chen did some great =
work updating the docker image. Check out the PR here.
revisited FCREPO-2938 - have folks written code (outside of fedora) tha=
t relies on the `<fedora:writeable>` triple? - it returns true no mat=
ter what and is an incorrect value which is misleading. =20
One solution that has support is removing `<fedora:writeable>` no=
w and always having the modify display in the Web Interface. So the i=
nterface will always look like you can edit a record, but will return an er=
ror if you can't.
Long term solution would be to put the ACL information into the header =
and have the Web Interface use that information when it decides whether or =
not to render the modify controls.
Peter Eichman will do t=
he first part by removing the <fedora:writeable> and create a ticket =
for the 2nd part. The 2nd part will not go into the 5.0 release.
Release date =E2=80=93 Danny Ber=
nstein will cut RC-2 on Monday the 19th of November. Release targe=
t would then be 3 weeks out on December 10th.