versioning and merging mechanics
2 views
Skip to first unread message
Chris Blow
Nov 25, 2009, 10:00:14 PM11/25/09
to talkin...@googlegroups.com
hello again talkinpapers-ers,
Mike and I have been talking about using git for some of the merging operations. We both love git for managing source code and are attracted to using it for an offline data sharing situation like this one. We have considered it primarily for it's merge command and it's native offline editing.
http://git-scm.com/
But now I am wondering if we need it at all: Presumably we will already have a normal database on the USB stick.
In my understanding, this stick will be the basis of the web application that allows users to create, update and print forms. The user in this case is offline but can print on a local network-in-a-tent. This person is scanning forms as they are received, and updating the forms for the next day's batch of surveys.
As I see it, the scanned data needs to be merged at two points:
1.) when it comes back to the network-in-a-tent, and is scanned into the locally running talkingpapers web app.
2.) when it is integrated with the internet, and with the surveys controlled by the parent organization.
is this right?
Merging two databases would then be something that could be handled in the web app layer (including offline mode). I'm imagining that the schema changes can be handled in a purely additive way -- the user can not take a destructive action on the data however much they PUT or POST changes. ... This would be the main benefit of representing the data in triplestore (or more realistically using a RDF API on the db). Triples are so easy to merge. Also, we don't need git to use hashes (globally unique IDs), we can use an ORM layer to handle that.
thoughts?
superbest
c
Mike and I have been talking about using git for some of the merging operations. We both love git for managing source code and are attracted to using it for an offline data sharing situation like this one. We have considered it primarily for it's merge command and it's native offline editing.
http://git-scm.com/
But now I am wondering if we need it at all: Presumably we will already have a normal database on the USB stick.
In my understanding, this stick will be the basis of the web application that allows users to create, update and print forms. The user in this case is offline but can print on a local network-in-a-tent. This person is scanning forms as they are received, and updating the forms for the next day's batch of surveys.
As I see it, the scanned data needs to be merged at two points:
1.) when it comes back to the network-in-a-tent, and is scanned into the locally running talkingpapers web app.
2.) when it is integrated with the internet, and with the surveys controlled by the parent organization.
is this right?
Merging two databases would then be something that could be handled in the web app layer (including offline mode). I'm imagining that the schema changes can be handled in a purely additive way -- the user can not take a destructive action on the data however much they PUT or POST changes. ... This would be the main benefit of representing the data in triplestore (or more realistically using a RDF API on the db). Triples are so easy to merge. Also, we don't need git to use hashes (globally unique IDs), we can use an ORM layer to handle that.
thoughts?
superbest
c
Reply all
Reply to author
Forward
0 new messages