| Time |
S |
Nick |
Message |
| 01:17 |
|
|
garnett joined #dvn |
| 13:45 |
|
pdurbin |
skay: we had a meeting about dataset versioning yesterday. I took some notes here if you're interested: https://docs.google.com/document/d/12W0YD1SGDmR1H0SpiQqFhTea7Ej-nAeSheEnePJ9xAs/edit?usp=sharing |
| 13:46 |
|
pdurbin |
yoh_: you might be interested in this too |
| 14:02 |
|
|
iqlogbot joined #dvn |
| 14:02 |
|
|
Topic for #dvn is now http://thedata.org - The Dataverse Network Project | logs at http://irclog.iq.harvard.edu/dvn/today |
| 14:05 |
|
|
javaeebot joined #dvn |
| 16:29 |
|
yoh_ |
pdurbin: I do not know much about the dataverse unfortunately so can't really provide adequate feedback. BUT deterministic versioning is indeed very important and great that you are looking it. |
| 16:32 |
|
yoh_ |
I guess I should check more on how dataverse organizes and exposes data and meta-data. ... major.minor to discriminate among data and meta-data changes also sounds good. I wonder if there might be any usefulness in e.g. skipping minor .0 for a new major. release if meta-data also gets changed with it (besides the major. number) |
| 16:32 |
|
yoh_ |
this way you would also know from the version that there was a change in meta-data |
| 16:48 |
|
pdurbin |
yoh_: apart from some universal basic rules about when minor or major versions must be bumped, we want dataset authors to be in control of the version numbers. for example, if a significant change to metadata justifies a bump to the major version, that's fine |
| 17:05 |
|
yoh_ |
makes sense ;-) |
| 17:07 |
|
yoh_ |
BTW -- are there are any datasets in the dataverse which do not demand explicit agreeing to terms of service? |
| 17:29 |
|
pdurbin |
yoh_: you know, I'm not sure... I don't know a lot about how terms of service have been implemented. I know for some data at http://thedata.harvard.edu you need to check a box and click an "I agree" button. I'd be curious to hear what your experience has been. Also, other Dataverse Network installations may have different rules. I'm really not sure. |
| 18:01 |
|
yoh_ |
;-) and that was my question -- either there is any data which is available without clicking on any 'I agree'... my experience, if you had used any linux distribution -- imagine clicking 'I agree' for every package ;) |
| 18:23 |
|
pdurbin |
yeah, that would be annoying. a lot of... friction |
| 18:23 |
|
pdurbin |
yoh_: please check out this related ticket: Feature #3277: Data Sharing API: Allow data owner to download without TOU. - Dataverse Network - Project Management at IQSS, Harvard University - https://redmine.hmdc.harvard.edu/issues/3277 |
| 19:23 |
|
|
garnett joined #dvn |
| 22:25 |
|
|
axfelix joined #dvn |
| 22:47 |
|
yoh_ |
interesting -- thanks! well... interesting fact that there might be no such a thing as "data owner" since data itself are not copyrightable in many jurisdictions... what is copyrightable is "collections of data" or "datasets" composing multiple pieces of data |
| 22:49 |
|
pdurbin |
yoh_: does the phrase "Open Data license" mean anything to you? I just saw that mentioned in an internal list |
| 22:49 |
|
yoh_ |
also I guess, in some cases where such agreement is stictly necessary (e.g. privacy conerns) it could be mitigated if there was a registry of 'access keys'. e.g. I would agree to some generic TOU, obtain my "key" and use it for download later on |
| 22:49 |
|
pdurbin |
javaeebot: lucky open data license |
| 22:49 |
|
javaeebot |
pdurbin: http://opendatacommons.org/licenses/odbl/ |
| 22:49 |
|
pdurbin |
huh. that's for databases, apparently |
| 22:50 |
|
yoh_ |
;-) read those "open data license" terms... e.g. on http://opendatacommons.org/licenses/ -- it is either "public domain dedication license" or "Database license" |
| 22:51 |
|
pdurbin |
oh, I see... "Public Domain for data/databases" |
| 22:51 |
|
pdurbin |
maybe this is the right website then |
| 22:52 |
|
pdurbin |
seems to be an http://okfn.org project |
| 22:52 |
|
yoh_ |
and IIRC "database" adds some interesting sides to it (so it is not just data or a dataset... but I could be wrong) |
| 22:53 |
|
pdurbin |
yoh_: and then the dataset author could revoke your key? if there was reason to believe it had been compromised, for example? |
| 22:57 |
|
pdurbin |
not that a single password is any more secure. could also get compromised |
