Date: Thu, 28 Mar 2024 04:36:02 -0400 (EDT)
Message-ID: <1080784974.27268.1711614962024@lyrasis1-roc-mp1>
Subject: Exported From Confluence
MIME-Version: 1.0
Content-Type: multipart/related;
boundary="----=_Part_27267_1198830863.1711614962024"
------=_Part_27267_1198830863.1711614962024
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
Content-Location: file:///C:/exported.html
2017-07-26 - API Spec Meeting
2017-07-26 - API Spec Meeting
Time/Place
- Time: 2:30pm Eastern Time US
- Dial-in Number: (712) 775-7035=20
Attendees
(facilitat=
or)
(notet=
aker)
Agenda
- Open PRs ...I believe the=
y are all ready to go
- Targeting "all 5 sections covered by end of Aug"=20
- Resource Management
- Resource Versioning
- Resource Authorization (by week of July 24)
- Notifications
- Binary Resource Fixity
- Open Authorization issues
- Re: #168: Referencing ACLs/grou=
ps outside of the repository =E2=80=94 we should add a non-normative note a=
bout this=20
- Maybe add an item to the security considerations section too?
- Re: #176:ACL creation and linki=
ng -- be explicitly silent or specify?
- Specifying the expected relationships =20
- acl:accessControl from protected resource, and
- acl:accessTo or acl:accessToClass from ACL, and
- Link: <some-acl>; rel=3D"acl" from protected resource
- Re: #175: Should way say d=
efault ACL MUST be resolvable?=20
- Yes
- Re: #172: Clarify algorith=
m for finding authorizations=20
- Pending response from Solid
- Re: #170: Require acl:Appe=
nd and acl:Control=20
- The suggestion is to be in line with the Solid spec on this point... th=
erefore, Fedora spec requires no additional text.
- Re: #165: Removing support=
for acl:accessToClass?=20
- This is a very useful construct
- Re: #164: Requiring acl:de=
faultForNew?=20
- Pending response from Solid
- Re: #134: Restore ACL lang=
uage to HTTP DELETE with Depth header
- To be resolved with: https://git=
hub.com/fcrepo/fcrepo-specification/pull/178
- Looking good on Deletion-related issues?
- Test suite contract is underway=20
- Should be able to close: Selec=
t language for development of Fedora API test suite
- Fedora API Del=
ta Document in-progress
Minutes
- Daniel Lamb is to review #181, and #182 i=
s ready to follow up afterwards. #180 and #178 are waiting Benjamin Armintor
- Resource versioning has been identified as the next topic to tackle to =
help us meet our deadline of end of August, and S=
imeon Warner has volunteered to be point man on the topic. Everyo=
ne has homework to read the Memento spec and respond to issues Simeon raise=
s on Github. Esm=C3=A9 Cowles will take a loo=
k at the fixity section in parallel.
- Open issues:=20
- Labeled as resolved-needs-pr for some non-normative language.
- Current thoughts are to allow a user to provide a link header with rel=
=3D'acl' to point to an already existing access control policy on PUT or PO=
ST. If the header is not preset, default behavior is left to the impl=
emntor and can be either 'do nothing' or 'create a new policy'.
- Yes, we feel making it resolvable is appropriate.
- Still pending a response from SOLID
- It could be challenging to filter through PATCHes or deltas on PUTs, bu=
t it could be navigated. Possibly we could require it for PATCH only,=
and leave PUT unsecified. Agreement on what we mean by 'append' woul=
d be useful to include in the specification (e.g. no PUT, no DELETE, no PAT=
CH with DELETE statements).
- Consensus seems to be around removing it, and it's something that can b=
e highlighted for people's consideration if anyone actually needs it.
- The SOLID team will be adding 'default' to the vocabulary.
- No concerns have been raised and the deletion issues should be wrapped =
up soon.
- Looking good.
Action Items
-
------=_Part_27267_1198830863.1711614962024--