Time
S
Nick
Message
01:23 donsizemore joined #dataverse
01:44 djbrooke joined #dataverse
01:54 nico___ joined #dataverse
03:07 iamtimmo joined #dataverse
04:50 soU joined #dataverse
06:41 axfelix joined #dataverse
06:44 axfelix joined #dataverse
07:40 axfelix joined #dataverse
08:17 axfelix joined #dataverse
12:51 donsizemore joined #dataverse
13:18 andrewSC joined #dataverse
14:13 iamtimmo joined #dataverse
14:28 pameyer joined #dataverse
14:30 andrewSC
morning all
14:31 pameyer
morning
14:36 andrewSC
question: When I publish a dataset, does anything create or publish links to that dataset outside of my dataverse? Basically I'm trying to figure out how the DOI system works. My Dataverse install is behind a proxy for org. reasons and we don't want any public links being made to our datasets
14:37 andrewSC
is the DOI basically just a hash of the dataset?
14:44 andrewSC
hmm judging by the demo dataverse it seems like the doi links aren't valid/working on published datasets anyways?
14:50 nico___ joined #dataverse
15:02 soU joined #dataverse
15:19 djbrooke joined #dataverse
15:50 djbrooke
Hey andrewSC - I think I remember a discussion a ways back about this
15:50 andrewSC
cool
15:52 djbrooke
DOIs in the application are very important to the whole sharing/publishing thing so I think you'd have to creatively break a few things to accomplish what you want
15:53 andrewSC
djbrooke: I've modified the source to not display the DOI link itself, but I'm wondering, when I hit the publish dataset button, is anything "calling out" to other services on the internet to "register" the DOI or other links?
15:54 djbrooke
So yes, an outside link is generated, and there's not any promises that one of the DOI providers (EZID, Datacite) will not index/catalog/make available that DOI
15:54 andrewSC
For example, I know the dataverse site itself has a map of dataverse instances. Is that something automated/when I publish my first dataset publicly, the map is updated to display a new dataverse? Or is showing up on that map something that requires coordination with your team?
15:55 djbrooke
For Dataverse installations, showing up on the map requires manual intervention by us (and we'd love to add you guys!)
15:56 andrewSC
djbrooke: gotcha gotcha
15:56 djbrooke
For dataverses within harvard.dataverse.edu, if you click on the Harvard dataverse you'd see a bunch of smaller dots that represent the dataverses within harvard.dataverse.edu
15:56 djbrooke
err dataverse.harvard.edu
15:56 andrewSC
lol
15:56 djbrooke
(can't blame a typo because I did it twice)
15:57 andrewSC
;)
16:10 pameyer
andrewSC: regarding "calling out", I *think* both "create dataset" and "publish" make API calls to the DOI provider
16:10 pameyer
out of the box, this uses demo credentials that create temporary/test identifiers
16:11 andrewSC
ahhh gotcha
16:12 pameyer
if it's behind a proxy that blocks outgoing by default, then those should be stopped
16:18 andrewSC
pameyer: mhmm, I'm not seeing any resuts in ezid for the DOI's that have been created for the unpublished datasets
16:18 andrewSC
I'm fairly certain the proxy is in place for things going in, but not going out
16:19 pameyer
how are you querying?
16:26 pameyer
assuming your on the defaults, does curl http://ezid.cdlib.org/id/doi:10.5072/FK2/$foo show anything?
16:27 pameyer
for values of $foo where it matches a dataset in your install
16:34 iamtimmo joined #dataverse
16:44 iamtimmo joined #dataverse
17:01 andrewSC
pameyer: just got back to my desk, I was querying by going to the http://ezid.cdlib.org/search and pasting the DOI from a doc in the identifier field. Eating part of my lunch now though, lemme check the curl in a sec
17:16 axfelix joined #dataverse
17:25 djbrooke joined #dataverse
17:28 pameyer
andrewSC: no rush from my end (I'm about to head out for lunch too)
17:37 djbrooke joined #dataverse
17:38 djbrooke joined #dataverse
17:51 axfelix joined #dataverse
18:24 djbrooke joined #dataverse
18:27 andrewSC
pameyer: oh boy, it exists with the curl...
18:27 andrewSC
how do we remove this from cdlib
18:27 pameyer
it shows "_owner: apitest"?
18:27 andrewSC
yes
18:28 pameyer
it'll go away on its own - not sure of the timeframe, but the ezid API docs should have the specifics
18:28 andrewSC
pameyer: I can query you with a private gist if you'd like
18:28 andrewSC
gotcha
18:28 pameyer
10.5072/FK2 is a test shoulder
18:28 andrewSC
ok
18:28 pameyer
if there's info that you need to clear out manually, you can update the identifiers to overwrite
18:30 pameyer
http://ezid.cdlib.org/doc/apidoc.html - EZID deletes them after 2 weeks
18:30 andrewSC
gotcha gotcha, thanks for the heads up
18:31 pameyer
no problem
18:32 andrewSC
pameyer: If I put in something like "localhost" for doi.baseurlstring, would that effectively stop the DOI's from being generated? Or would that just cause errors/not allow people to create datasets?
18:36 djbrooke joined #dataverse
18:38 pameyer
it'll stop the DOIs from being externally registered; they're still being "created" by the dataset identifier and configured DOI shoulder
18:38 andrewSC
gotcha
18:38 pameyer
I could see it running into problems if the API call to localhost fails; but I'm not sure how that'll play out
18:39 andrewSC
yeah I'm fine with dataverse creating DOIs but I can't have anything external being notified of it if that makes sense
18:39 pameyer
most of my DOI/EZID familiarity is in a non-dataverse system
18:39 pameyer
it does make sense
18:39 djbrooke joined #dataverse
18:39 andrewSC
cool cool
18:40 pameyer
I've been vaguely annoyed for years now about my inability to figure out how to limit outgoing connections from a linux box
18:40 andrewSC
lol
18:40 pameyer
… without taking down the network entirely, I mean
18:40 andrewSC
hahah
18:40 andrewSC
touche
18:40 andrewSC
yeah it's pretty clear on how to stop things from coming in, but blocking stuff going out is not something I come across very often myself
18:41 pameyer
it seems pretty uncommon
18:47 andrewSC
oh interesting
18:48 andrewSC
so I made the change to the doi.baseurlstring setting to "http://localhost ", confirmed that the curl wasn't returning anything for the new dataset DOI I just made, but when I go to click publish I get "This dataset may not be published because it has not been registered"
18:48 andrewSC
I assume registered in this context is referencing not having a registered DOI?
18:49 pameyer
I think so
18:49 andrewSC
now that one is gonna be tricky haha
18:50 pameyer
at least moderately tricky - but it sounds like it would need a check disabled
18:51 andrewSC
hmmmm
18:51 pameyer
I think this was some of the "creative breakage" djbrooke was referring to :)
18:51 andrewSC
ahhh gotcha
18:53 pameyer
you're building your own war, right?
18:53 andrewSC
I don't think so... I think the last time I made changes (removing the DOI links from the view) I just re-compiled the java files on the server that I touched
18:54 andrewSC
I really should get that under some sort of VCS
18:55 pameyer
VCS is almost always a good idea
19:01 djbrooke joined #dataverse
19:04 andrewSC
yeah... At this point I'm kinda boned since I don't remember the changes I made.. I can however make this "feature" change in a fork and start from there..
19:04 andrewSC
guh
19:16 djbrooke joined #dataverse
19:28 andrewSC
pameyer: any suggestions on where I could start spelunking for that change?
19:29 andrewSC
looking at the network tab after clicking publish leads me to dataset.xhtml
19:33 pameyer
I'd start with following the trail down the publish dataset command (java, not xhtml)
19:35 andrewSC
lol there is literally something called "PublishDatasetCommand". Nice :D
19:37 pameyer
yup
19:37 pameyer
;)
20:21 joshb-ori joined #dataverse
20:24 joshb-ori left #dataverse
20:52 djbrooke joined #dataverse
21:24 andrewSC joined #dataverse
22:02 sivoais joined #dataverse
22:12 sivoais joined #dataverse
22:41 djbrooke joined #dataverse
22:44 djbrooke joined #dataverse
23:13 jeffspies______ joined #dataverse
23:49 nico___ joined #dataverse
