Make it easier to ‘join the club’ of participating systems in the federation
Angus: we’ve moved on since the start of the project from saying we try to make just one format
Hence the ‘club’ model, which is about not having the same format but still having some conventions, to spread the load
Angus: action to elaborate on the journey
George: we weren't aware of that one specifically, but should be feasible.
Michiel: we can just add that to PreJournal.
Victor: we can do CalDav format
It's not a union format, but does support a pretty broad range of data in timesheets, as well as metadata.
George: we could potentially include CalDav explicitly in all our systems, as part of the milestones.
George: is there such a thing as a CalDav endpoint?
Victor: it's a protocol, supporting events.
Michiel: iCal is the data format.
Michiel: happy to do push of CalDav data on behalf of the rest of the federating systems.
Michiel: organisations are separate from software code bases, so you could run your own instance of PreJournal to integrate with say, Google Calendar.
Angus: does this mean that you want to go beyond just having data pushed to PreJournal?
Michiel: yes, using a GET against your endpoint to list the data in question.
George: if we do that, we potentially lose the benefit of reducing the work involved with Federation by halving the effort to just pushing.
Michiel: if you only push, there may be data loss and /or corruption, so being able to do a pull enables that to be detected and addressed.
Michiel: if a given system relies on doing a "push-and-forget", then it wouldn't have the authoritative copy of the original data.
Angus: the source system would still retain the original data, but would benefit of only having to push data, and not read it.
George: yes, but they don't have to know how to read them, only write them.
When a new system joins the club, it has to know how to post data to other club members, and those members have to learn its target format for it to be included in the club.
Michiel: it's almost about content negotiation, since the format could be expressed as a MIME Type.
George: that would be a GET, though, rather than a POST.
Michiel: what happens if the number of systems in a federation grows significantly beyond just the 3 that are in it now?
Angus: it wouldn't be sustainable to continue connecting as a mesh if there are more than 5 or 6 systems; it would be better if some act as proxies for others, essentially creating "islands of federation".
Michiel: we could simulate that by each of us integrating with only one other system in this project?
Angus: we could potentially compromise the objectives of this project if we do that; it would be better to maintain the 3-way federation we are currently aiming at, with a view to proposing a follow-on project that shows how this would scale to a larger number of systems, with some proxy arrangement as described above. The bigger picture is the strategic goal of supporting genuinely federated bookkeeping, where the number of systems and system instances needing to be connected would be legion. The Internet itself is a good precursor of this, in that its IP layer supports such islands, using protocols such as BGP, and concepts like Autonomous Systems.
Michiel: OK, this would help us to justify such a project.
Michiel: the original idea was that Andrei would be writing that up on Ponder Source's behalf, but he's been so busy with his primary job that he's not been able to take that on. Rose will be doing some of it, however.
George: Angus has an action to describe the Information Architecture journey from the original conception to what it is today.
Angus: yes, it needs to articulate how we moved from a union format for the data to more of an autonomous, decoupled, yet connected approach, manifest in the "POST-only in target system's format" line we are now taking.
Michiel: would it make sense for that milestone to be shared between Ponder Source and m-ld, in view of that work Angus has?
Michiel: OK, so we'll ensure we track time on that task.
Angus: and we'll do the same on our side.
@kroky_gitlab https://timesheets.dev3.evoludata.com/tiki-index.php has an incorrect TLS cert and is showing "Ceci est un test privé pour le CPMD.", is that intended?
1 reply
@kroky_gitlab when you say you support Scoro, Mite and Time (https://gitter.im/federatedbookkeeping/timesheets?at=62ab2467e393a318065d4dd1), do you mean there is a way to sync data from those into WikiSuite, out of WikiSuite to those, or both ways? And what is the process for each? Can it be unattended or does it involve the user downloading + uploading a file?
1 reply
prejournal=# select c.name, sum(m.amount) as hours from movements m inner join components c on m.fromcomponent = c.id where m.tocomponent=8 and m.timestamp_ < date('2022-08-05 00:00:00') group by c.name;
name | hours
---------+-------
ismoil | 476
michiel | 221.5
(2 rows)You should be seeing the same numbers when you query the API of https://time.pondersource.com
This does mean that we're already over budget, especially if we give half of milestone 4 to m-ld as discussed on Monday's meeting. That's OK because the Ponder Source Foundation can finance the rest, but it does mean we're nearing wrap-up time from our side. :) Ponder Source Foundation also financed the hours made by Triantafullenia and Lambert already, and Andrej worked as a volunteer on the project from our side.