Skip to content
This repository has been archived by the owner on Nov 12, 2019. It is now read-only.

Can't make partial changes to rows #4

Closed
Raynos opened this issue Aug 27, 2012 · 5 comments
Closed

Can't make partial changes to rows #4

Raynos opened this issue Aug 27, 2012 · 5 comments

Comments

@Raynos
Copy link
Contributor

Raynos commented Aug 27, 2012

Consider peer X and peer Y which each have a row.

Now let peer X do row.set("one", "one") and peer Y do row.set("two", "two")

This propagates messages [row.id, { one: one }, ts1, source] and [row.id, { two: two }, ts2, source]

For normative purposes let ts1 < ts2

Now if any peer gets these messages in order of ts2 followed by ts1 then we lose the set("one", "one") message because it's older and ignored.

@Raynos
Copy link
Contributor Author

Raynos commented Aug 27, 2012

I think this is simply a misunderstanding of how crdt works

@Raynos Raynos closed this as completed Aug 27, 2012
@dominictarr
Copy link
Owner

That is a bug.
Each source gets it's own timer.
A message should only be ignored if there is a more recent message for that source

@dominictarr dominictarr reopened this Aug 27, 2012
@Raynos
Copy link
Contributor Author

Raynos commented Aug 27, 2012

@dominictarr does that mean that if we send crdt messages over UDP then messages can arrive in the wrong order and not get applied?

@dominictarr
Copy link
Owner

yeah, scuttlebutt depends on correct ordering.

if you where using udp, I'm assuming for an action game, you might have
different assumptions. maybe just send the whole row, or something.

On Mon, Aug 27, 2012 at 9:49 AM, Raynos notifications@github.com wrote:

@dominictarr https://github.com/dominictarr does that mean that if we
send crdt messages over UDP then messages can arrive in the wrong order and
not get applied?


Reply to this email directly or view it on GitHubhttps://github.com//issues/4#issuecomment-8049089.

@Raynos
Copy link
Contributor Author

Raynos commented Sep 1, 2012

@dominictarr I'm going to assume this is a non issue because I havn't reproduced it recently.

@Raynos Raynos closed this as completed Sep 1, 2012
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants