Even dev code should build in my opinion.
And patience you say… what’s keeping the safe network from becoming something like GNU Hurd,
where at some point a single programmer takes the whole project, simplifies one part and then makes a release?
We would all love to see that, the more folk the better. No point in releasing an anti-network though, best to release the real network and use the real design/principles etc.
I can assure you 100% normal users will not care about master branch and we had a broken master. Parsec was dropped, master was parsec. If you are desperate for that then it’s tagged and working in master and you can build that, it will kinda run etc. but it’s not Safe it’s another design.
4 Likes
Just for the records (I’m sure some part of future history will be grateful for this) the last combination of versions that worked together (using parsec) are
safe_vault v0.25.1 commit:5d5c214deb9b48dbb6ec532ca13e39a338eb0eff
safe_api v0.15.0 commit:0feadec0d0c196f57458b916d1c9b491e396906b
routing v0.37.0 commit:97cc0cdb6acf073fb53b9f679181176005db1f2b
quic-p2p v0.7.0 commit:a1ed4c4284445194caad7728f065ccc851c4e3ba
safe-nd v0.10.1 commit:69dc14125413ac17ae36190e462b8a488be4be9d
safe_core v0.42.1 commit:9ff407420a1ef577c545d60881a9dea44488ed83
threshold_crypto v0.3.2 commit:485333db6e2611d26a2a921a6ee1e1dcc3ce4624
BLS-DKG v0.1.0 commit:be6ac3e3eb0835086f17b694f551118fa8bb0de8
parsec v0.7.1 commit:b71dfb3b8843f647c853fdc6925f670d67024cb7
@folaht I am aware you are looking for something other than the versions above. You are looking for something that is not yet ready (Safe Network on LAN), and much as I wish I could say ‘do this and this and that’ the truth is right now what you are asking for does not exist.
Is there any way for you to move toward this goal in parallel while the network is under development? Maybe a mock network so sites can still be uploaded and then once Safe Network is where you need it you can easily migrate across? Just hoping to get you started on the things you see being valuable if possible…? How important is it to be a ‘real’ Safe Network instance? Would a mock be ok? I would imagine a day of solid coding could produce a suitable mock API + client for the network.
6 Likes
Another vector/aside here is the move we are currently on. As we get the types CRDT compliant we have offline working. This is where a client can continue data objects offline and sync them again.
I am gonna do a presentation on this soon, in house. Anyway here’s a synopsis.
A data container (any of them) has an owner. The owner (Actor, could be many owners) signs each operation on the data (add item etc.) and the crdt is valid, i.e. if it merges locally it will merge everywhere. So all you need is a “bag of containers” to mock out the system.
Actually though that is only Actor valid data, this is normally all CRDT types need to be valid, however we have “network Valid” actions as well. So this is a bit more complex (but not much). It works like this.
Client “pays” for updates to the data, this could be 1 or many operations. On payment the network (keep it simple) signs the data at that point in causal time is network valid and paid. The data type can drop any other network payment records in the data as the latest payment is only valid if all previous operations are paid for.
This allows a client to merge cleanly back to the network. It’s based on AT2 and can make our lives so much simpler, but requires the client (Actor) does store the paid for data. This part is still a concern for us and is solvable (i.e. the network takes the paid for data and stores on paid for) but it would be great if the network did not store on behalf of the client as it’s much less work (and saves us committing a transaction to put it back in the clients hands (AT2).
In any case, offline working should get considerably simpler, mock to as it can use real data and real signatures.
7 Likes
This is a really great summation of how all the work you’ve done is going to make this cleaner and easier.
It’s obviously secondary to getting the actual network up and running, but I do think that once that happens (along with the inevitable restarts, fixes etc.) that having a way for new blood to come in and develop immediately will be critical to maintaining momentum. Right now there is a small cadre of skilled devs who have put up apps on previous iterations, but this group will need to expand. Supporting them so anyone can find out about SAFE and start building even between network restarts is of extremely high importance.
3 Likes
I’m sure that “don’t break the build” may become more of a mantra in beta. As you saw from other comments, a lot of radical design changes are inbound so no point right now in trying to get old working test versions up and running.
4 Likes
@StephenC won’t let us and all the recent ci/cd work is to enforce clean builds and releases. This is why a few crates are currently held back (routing/node etc.) as they stabilize. When they stabilize the build will continually work.
13 Likes
I could simply build it. But it’s just that the ‘feel’, the ‘satisfaction’, isn’t there.
It’s like designing 3D models for 3D printing without being able to print them.
Even a LAN gives a bit of an illusion.
Besides, my computer does not have a lot of memory.
I prefer running 8 nodes on another computer for that.
Okay, I’ll do it despite not being able to have others see my results.
A personal site and a pet project?
Not much I would gather.
I guess so…
3 Likes