couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gregor Martynus <gre...@martynus.net>
Subject Re: using POST _bulk_docs with ?edits=false
Date Fri, 24 May 2013 11:04:06 GMT
Thanks Jason, I've a better understanding of the ?edits=false flag now and also what the cause
of my problem is. After looking more into it, I've found that conflicts are not created on
every POST /_bulk_docs?edits=false, but only for some documents. And the reason is that the
`_revisions.ids` history is not correct in all cases, so it must be a Hoodie bug.

PS: Special thanks that you found the time to respond in these exciting times :-) Very happy
for you & IrisCouch 

-- 
Gregor Martynus


On Friday, 24. May 2013 at 07:36, Jason Smith wrote:

> Calculating _rev "correctly" depends on what you need.
> 
> If you and I have the same document and make the same changes except _rev
> (and _revisions), then we will get a conflict when we replicate to each
> other.
> 
> The only project I know that uses edits=false somewhat successfully is
> mikeal/replicate. It might help, although it gets its _revisions value from
> CouchDB, it doesn't calculate anything.
> 
> Can you paste some example documents (with the _revisions) value?
> 
> 
> On Fri, May 24, 2013 at 3:28 AM, Gregor Martynus <gregor@martynus.net (mailto:gregor@martynus.net)>wrote:
> 
> > Hey,
> > 
> > I'm using the edits=false flag to push changes from the browser to the
> > couch, to prevent that data can get lost. My current implementations
> > randomly generates a _rev number and adds the _revisions attribute with the
> > current _rev and the newly generated _rev number. But that leads to
> > conflicts. I was told that the _rev number has to be calculated based on
> > the objects.
> > 
> > I try to find out how to calculate a new the _rev number correctly but
> > can't figure it out. I've described the issues here:
> > https://github.com/hoodiehq/hoodie.js/issues/66
> > 
> > Would really appreciate your help
> > 
> > --
> > Gregor
> > 
> 
> 
> 



Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message