Time
S
Nick
Message
07:03 jri joined #dataverse
07:17 jri joined #dataverse
08:08 jri joined #dataverse
08:53 stefankasberger joined #dataverse
09:01 poikilotherm joined #dataverse
09:47 pdurbin joined #dataverse
09:55 pdurbin
jri: thanks for creating https://github.com/IQSS/dataverse/pull/5995 (and the issue)! I just moved it to QA.
09:56 jri
Hi :) no problem!
09:58 pdurbin
How's the upgrade going? :)
09:58 poikilotherm
Mornin' pdurbin :-)
09:58 poikilotherm
How where the fireworks?
09:59 poikilotherm
Bingewatched Stranger Things Season 3 yesterday with 4th of July fireworks :-D
09:59 pdurbin
Exciting news. You can see them from one of the bedroom windows of our new house. Not very well, through the trees, but at least you can see them. :)
09:59 poikilotherm
Great! :-)
10:00 pdurbin
They don't start until 10:30pm, which is a little late for the kids. And for me, honestly. :)
10:01 poikilotherm
Yeah, early birds tend to fall of the tree when the bang and boom starts ;-)
10:01 poikilotherm
Although I heard that birds can attach themselfs to trees with their claws...
10:03 poikilotherm
I dunno if you guys use Logitech hardware... If you do, be warned... https://www.heise.de/ct/artikel/Logitech-keyboards-and-mice-vulnerable-to-extensive-cyber-attacks-4464533.html
10:04 pdurbin
Yikes. Look out for 2015 MacBook Pros too. I have one but when I entered my serial number I was told I'm not affected: https://petapixel.com/2019/07/03/apples-macbook-pro-battery-recall-is-no-joke-heres-what-could-happen/
10:05 poikilotherm
Ouchn.
10:06 poikilotherm
We have an old notebook as an exhibit downstairs... The charger catched fire a few years ago during the night hours. Don't mess with LiPo cells...
10:08 pdurbin
:)
10:16 pdurbin
poikilotherm: heh, I like the column at https://github.com/orgs/IQSS/projects/4#card-23433534 called "I'll be back!"
10:16 pdurbin
stefankasberger: wasn't that your line? :)
10:16 poikilotherm
:-D
10:17 pdurbin
or maybe that was some other Austrian
10:30 pdurbin
poikilotherm: did you see Slava put his talk from #dataverse2019 on YouTube? https://www.youtube.com/watch?v=vAPpKuDQUDY
10:30 poikilotherm
Nope
10:32 pdurbin
via https://twitter.com/4tykhonov/status/1146805688720592897
10:35 poikilotherm
Heh. "It was written by some German folks, so it is very strict"... :-D
10:36 pdurbin
that got some laughs :)
10:40 poikilotherm
Yeah... ;-)
10:50 pdurbin
poikilotherm: I just left you a comment at https://github.com/IQSS/dataverse/issues/5991#issuecomment-509177309
10:50 poikilotherm
Thx
10:50 poikilotherm
I saw that other issue, too
10:51 poikilotherm
I dunno if this might help
10:51 poikilotherm
It can, but that depends on the other peoples use cases
10:52 poikilotherm
And #4383 was already mentioned in the description :-D
10:53 pdurbin
huh, sure enough :)
10:56 pdurbin
poikilotherm: I just left a comment here as well: https://github.com/IQSS/dataverse/pull/5997#issuecomment-509178933
10:57 poikilotherm
I re-commented in #5991 :-)
10:57 pdurbin
Thanks. I'm going through unread email in order but I'll skip ahead. :)
10:58 pdurbin
ok, so it might help
10:59 pdurbin
I guess I'm still a little sad that https://github.com/IQSS/dataverse/pull/4341 was closed. It meant another fork.
11:00 pdurbin
huh... maybe not... https://dataverse.no seems to be using SAML rather than OAuth... so I guess they didn't fork
11:01 pdurbin
That's the thing, the OAuth providers are hard coded. Our SAML implementation allows for more flexibility.
11:01 poikilotherm
Meh. Only as long as you are using Shib.
11:01 pdurbin
It would be great if there could be similar flexibility with OAuth.
11:02 pdurbin
I bet your users won't care if you're using shib or something else. :)
11:03 poikilotherm
Sure.
11:03 poikilotherm
Admins care
11:04 poikilotherm
We can let others run the service for this
11:04 pdurbin
ah
11:04 poikilotherm
We plan to use https://login.helmholtz-data-federation.de
11:04 poikilotherm
This service is run by others, uses Unity IDM
11:04 poikilotherm
We will just need to register as an OAuth/OIDC client
11:05 poikilotherm
We will get all the nice stuff for free... It has groups support, offers attributes from the SAML world etc
11:05 pdurbin
Do they support SAML?
11:05 poikilotherm
Sure
11:06 poikilotherm
The list of institutions on the right is bases on eduGain
11:06 pdurbin
Ok, but you don't want to run mod_shib and shibd.
11:06 poikilotherm
You could login with your Harvard account
11:06 poikilotherm
We don't need to
11:06 poikilotherm
Just reuse existing services
11:07 poikilotherm
Less admin work needed
11:07 poikilotherm
No need to run the SP service as with Shibboleth
11:09 poikilotherm
pdurbin I just left a comment at https://github.com/IQSS/dataverse/pull/5997#issuecomment-509180197
11:09 poikilotherm
Please let me know how much details you guys want/need
11:11 pdurbin
Thanks. I think it comes down to cost/benefit analysis. What do we get out of the code? Sounds like a new version of ScribeJava. What is the cost? The time testing it.
11:11 poikilotherm
Which is a good idea anyway :-D
11:12 pdurbin
The ScribeJava people have been busy, I guess.
11:12 poikilotherm
People = 1 guy ;-)
11:13 poikilotherm
https://github.com/scribejava/scribejava/graphs/contributors
11:13 pdurbin
I can understand not wanting to run an SP, I guess but I think it's a situation where you "set it and forget it". Once it's set up I think it pretty much just works.
11:14 poikilotherm
Heh. Maybe in your AAIs
11:15 poikilotherm
We have plenty of people in eduGain coming from IdPs not sending proper attributes
11:15 poikilotherm
Learned that via our Gitlab deployment
11:15 poikilotherm
And admins "on the other side" tend to have not much time for fixing these kind of problems
11:16 poikilotherm
That makes a bad UX
11:20 pdurbin
Ah, good point.
11:22 pdurbin
For a long time we've wanted our SAML implementation to work more like our (newer) OAuth implementation and simply prompt the user to enter attributes that weren't asserted by the IdP.
11:23 poikilotherm
Yeah. On the other hand it might come in handy to use some other service Dataverse is attached to for realizing feature parity
11:23 poikilotherm
Shib support has groups
11:23 poikilotherm
OAuth does not
11:23 poikilotherm
OAuth has the nice input form for attributes missing
11:23 poikilotherm
Better handling of stuff etc
11:23 poikilotherm
Shib does not
11:24 poikilotherm
Maybe it is easier to let another service do the hard job of authn/z and then hand over to Dataverse
11:24 jri joined #dataverse
11:24 poikilotherm
With maximum flexibility for all
11:26 poikilotherm
I dunno what's right, this just my view of the world right now...
11:26 poikilotherm
+is
11:26 poikilotherm
I might be wrong, forget use cases and obviously I am not everybodys guru. I just like to have choices
11:36 poikilotherm
BTW pdurbin what do you think of https://github.com/IQSS/dataverse/issues/6000
11:36 poikilotherm
(I just noticed that no. 6k is mine :-D Ah, what an honour...)
11:37 poikilotherm
(You might be interested in adding another graph to your analysis, regarding the numbers of issues created and closed over time...)
11:37 poikilotherm
(GitHub doesn't have them... It might be interesting to see how the attention for the project raise. I think a year ago we were at ~5k)
11:45 jri joined #dataverse
11:46 pdurbin
The guy who opened issue 4000 tweeted about it: https://twitter.com/aecoppock/status/885439123112308736 :)
11:47 poikilotherm
Hehehe. I think of doing this, too :-D
11:48 pdurbin
poikilotherm: how much do you care about the user experience in Dataverse of SAML vs. OAuth? With OAuth you can decide your own username, your own email address, etc.
11:48 pdurbin
When signing up for Dataverse, I mean.
11:48 pdurbin
And with OAuth accounts you can change your email address at any time.
11:49 poikilotherm
Yeah. I won't bother too much about the user choices for such fields. Ideally, they are prefilled by an IDM anyway
11:49 pdurbin
ok
11:49 poikilotherm
I want the UX to be as smooth as possible
11:49 poikilotherm
People should not hang like with missing attributes
11:50 pdurbin
yeah
11:50 poikilotherm
This is really frustrating and will keep people from attending not only the service, but most likely reject RDM as a goal, too
11:50 poikilotherm
Errr...
11:50 poikilotherm
Other way round
11:50 pdurbin
I don't think we have an issue for this. How it hangs and fails.
11:50 poikilotherm
not attending
11:51 poikilotherm
Well I have just the XP from our Gitlab deployment
11:51 poikilotherm
I don't want to run into this in a field as RDM, where most scientists see things more as a burden than a pleasure
11:52 pdurbin
Sure. Smooth UX.
11:52 poikilotherm
At least here it is very often like this. I should mention here that we have no to just a few social science guys
11:53 poikilotherm
And our Dataverse is primarily aimed at being a metadata repo
11:53 pdurbin
No files? :)
11:53 poikilotherm
Scientist should use community repos, but we need a catalogue of everything going on
11:53 poikilotherm
We will try to harvest as much as possible
11:53 poikilotherm
If we can avoid it, yes, no files
11:53 poikilotherm
Just as a last resort
11:54 poikilotherm
Most community repos are most likely better at their metadata schemas and support for the necessities
11:55 poikilotherm
No files also due to the nature of many of our use cases, where datasets easily have TiB to PiB
11:55 poikilotherm
"Unmovable"
11:55 pdurbin
https://data.inra.fr has way more datasets than files, which is somewhat unusual for a Dataverse installation.
11:56 poikilotherm
We'll join 'em then ;-)
11:57 pdurbin
I guess my main question or thought on https://github.com/IQSS/dataverse/issues/6000 is "What is the longest endless checkbox nightmare currently shipping in Dataverse and can a screenshot of it be added to the issue?"
11:58 pdurbin
Also, if you write just enough code to demo a different approach, I'm happy to spin up the branch for UX people to look at and play with.
11:59 poikilotherm
Ok I'll see what I can come up with...
12:00 pdurbin
awesome
12:51 donsizemore joined #dataverse
14:06 andrewSC joined #dataverse
14:37 donsizemore
@pdurbin i've plunked out a Dockerfile for miniverse as-is, but am abandoning the pain of making it python36-compatible
14:39 pdurbin
donsizemore: wait, miniverse is dead. Long live dataverse-installations and the little Django app inside.
14:39 donsizemore
@pdurbin oh, poop. i thought you wanted to document miniverse installation so i've cobbled a Dockerfile around it. just dealing with venv and python3 issues
14:40 pdurbin
well, it's all Django so probably no harm done :)
14:40 pdurbin
Docker is more fun than Ansible?
14:40 donsizemore
lower cost of entry for somebody wanting to try out miniverse
14:40 pdurbin
The nice thing about Docker is that we could nudge devs toward it too.
14:41 donsizemore
clone the repo, populate local_settings.py and build the image
14:41 pdurbin
Sure, and I guess more and more people have Docker installed these days.
14:42 pdurbin
Only those three steps in a README sounds great. Here was my original thinking, my original need: https://github.com/IQSS/dataverse-installations/issues/1
14:42 pdurbin
I assume you have `python manage.py` somewhere in your Dockerfile.
14:43 donsizemore
due to Docker/Django/venv i'm adding a script
14:43 pdurbin
whatever works :)
14:43 pdurbin
Mostly I just want to make it easy for devs to hack on.
14:43 pdurbin
To start implementing some of the ideas from the community meeting.
14:44 pdurbin
"installations profiles" and what not
14:44 pdurbin
poikilotherm: like installations that don't have very many files. That's a profile. :)
14:44 pdurbin
or were they called "installation personas"
14:44 pdurbin
whatever
14:44 pdurbin
:)
14:53 pdurbin
donsizemore: since you're on Twitter: https://twitter.com/reddit_datasets/status/1147965911409651712
15:05 pdurbin
see also https://github.com/IQSS/dataverse/issues/5999#issuecomment-509259189 :)
16:29 donsizemore
@pdurbin i ain't read carefully enough. want me to stumble through Dockify-ing that repo, and document my steps in the README?
16:29 pdurbin
donsizemore: yes, please!
16:30 pdurbin
I have a couple devs in mind who might be able to hack on the code once there's a README for how to spin it up.
16:34 jri joined #dataverse
16:35 jri_ joined #dataverse
16:36 pdurbin
donsizemore: when you mentioned `free` at http://irclog.iq.harvard.edu/dataverse/2019-07-02#i_99670 I should have asked if you tried jconsole.
17:32 bricas_
pdurbin: hey -- i didn't even realize there were github ID numbers. interesting.
17:40 jri joined #dataverse
17:44 dataverse-user joined #dataverse
18:23 pdurbin
bricas_: heh. You're an early adopter, apparently. :)
19:25 pdurbin
and willing to read through a lot of message backlog :)
20:53 poikilotherm joined #dataverse
22:07 brett-soric joined #dataverse
22:56 brett-soric left #dataverse
23:18 jri joined #dataverse
