Time
S
Nick
Message
06:09 dataverse-user joined #dataverse
07:34 poikilotherm joined #dataverse
08:05 jri joined #dataverse
10:12 MrK joined #dataverse
11:16 juancorr joined #dataverse
12:45 poikilotherm
Hey @all :-)
12:47 MrK
hey
12:47 MrK
in slack annotation would work :D
12:47 poikilotherm
Yeah ;-)
12:47 poikilotherm
So it does in RocketChat, Mattermost and IIRC Gitter
12:54 donsizemore joined #dataverse
13:00 MrK
We could move irc to something like that.
13:01 poikilotherm
Ssssshhhhh. pdurbin is around...
13:02 poikilotherm
I think it would be a good paraphrazation naming it "it's complicated"
13:09 pdurbin_m joined #dataverse
13:10 pdurbin_m
The culture of transparency Gitter brings to chat is healthy for free software. IRC , XMPP, Slack, and almost every other chat system encourage decision making behind closed doors.
13:14 pdurbin_m
right?
13:21 MrK
I mean I'm up for whatever modern chat :P
13:23 poikilotherm
I wonder if you could have Slack recording bot just as you have for IRC
13:24 poikilotherm
Or other chat systems
13:24 poikilotherm
Because if it is just about public logs and you could work around that, it opens up more options
13:26 pdurbin_m
Yes, it's just about public logs for me.
13:31 poikilotherm
Well at least for Slack, there is an API
13:32 MrK
There was slack chat logger on github I remember but for some reason it was no longer maintained and stoped working
13:32 poikilotherm
https://github.com/dutchcoders/slackarchive
13:33 poikilotherm
That one?
13:33 MrK
lemme check there was conversasion on jvm-poland
13:33 MrK
conversation*
13:35 MrK
ahhh they talked about but they are not using it anymore on public slack since "RODO" but it's still working probably.
13:39 pdurbin_m
People who use Slack don't expect it to be publicly logged. You would be going against the Slack culture.
13:41 poikilotherm
Well if you name it "public channel" and pin a message like "beware everything in this channel will be logged publicly" that should be ok?
13:44 poikilotherm
pdurbin this SAML thing is giving me headaches anyway :-(
13:44 poikilotherm
No eppn attribute from samltest.id
14:07 * pdurbin
hands poikilotherm some pain killers
14:08 poikilotherm
Did I get it right that eppn = eduPersonPrincipalName ~= mail in InCommon?
14:08 pdurbin
yeah
14:09 pdurbin
oh, well
14:09 pdurbin
mail?
14:09 pdurbin
not necessarily
14:09 pdurbin
eppn from HarvardKey is not my email address. it's an opaque identifier
14:10 * poikilotherm
sighs...
14:11 poikilotherm
Well, I will use mail for now as a workaround :-/
14:11 pdurbin
From the Dataverse perspective, the eppn must be unique. That's the main requirement. And unchanging.
14:12 poikilotherm
Yeah
14:12 poikilotherm
Digged through the code already
14:12 pdurbin
dug
14:13 poikilotherm
Thx. Chances are high for many IdPs around in Europe/eduGain will not give you either eppn nor mail
14:14 pdurbin
In the future, I don't want mail to be required from SAML IdPs. However, if not eppn, some sort of unique persistent identifier will always be required.
14:14 pdurbin
(We don't require mail from OAuth IdPs.)
14:14 poikilotherm
That's why I really like ORCID
14:15 poikilotherm
That is a good scientific ID and it will not change
14:15 poikilotherm
But logging in via ORCID is most of the time a no-go... People are worried.
14:15 pdurbin
Yes, and when you log into Dataverse with ORCID, your ORCID ID is stored as the persistent identifier for you.
14:15 poikilotherm
For different reasons and things
14:16 pdurbin
People worry too much.
14:17 poikilotherm
I am actually thinking about convincing my colleague to simply allow our local IdP plus ORCID
14:17 poikilotherm
I don't think people will accept it if there is no local institutional login
14:18 pdurbin
Makes sense.
14:19 poikilotherm
And ORCID would make it much easier for people leaving the institute
14:19 poikilotherm
They can access the old data just by using their unchanged ORCID login
14:25 pdurbin
Yep.
14:26 pdurbin
I'm not sure why you're saying "unchanged".
14:27 poikilotherm
Well, they might have changed their details at ORCID, but their ID is the same
14:27 pdurbin
I assue you'd convert their institutional account to an ORCID account if they leave your institute.
14:27 pdurbin
assume*
14:27 poikilotherm
Yeah, that would be a good option!
14:27 poikilotherm
The docs are not very precise about that...
14:27 pdurbin
Well, you have to convert twice, unfortunately
14:28 poikilotherm
Would I need to convert shib to local and then local to ORCID?
14:28 pdurbin
yeah
14:28 * pdurbin
ducks
14:28 poikilotherm
Ok
14:28 poikilotherm
Not very elegant, but as long as it works...
14:29 pdurbin
What I want next is self-service. People should be able to say, "I'm leaving this institute. I'm going to log into Dataverse and convert myself to a local account. Then I'll go through the local to ORCID account conversion process."
14:29 poikilotherm
Hell yeah
14:29 pdurbin
Want to build this for us? :)
14:29 poikilotherm
That sounds like a really needed feature
14:29 poikilotherm
LOL
14:29 poikilotherm
I think there are some other things to tackle first :-D
14:30 poikilotherm
Sounds like a hackathon project!
14:31 pdurbin
I suggested adding a "Convert to Username/Email Account" button and workflow here: https://github.com/IQSS/dataverse/issues/4430
14:32 poikilotherm
Meeeh... How I hate this... I have the IdP entity successfully transfered to Dataverse, but eppn is not working
14:32 poikilotherm
And IRC does not have that vulcano smiley
14:35 pameyer joined #dataverse
14:36 pdurbin
Maybe we should create "dataversebot" to add some of the features you want. :)
14:36 pdurbin
It would keep iqlogbot company.
14:37 pameyer
shib -> orcid needing local accounts means folks can create built-in accounts, right?
14:37 pdurbin
pameyer: right
14:39 pameyer
the config to hide the built-in sign up page exists; but still opens up it up as an auth type. might or might not matter, depending on what the plan of a given installation is
14:39 pdurbin
yeah
14:40 poikilotherm
IF one could avoid the local accounts during conversion, that would be really awesome
14:40 poikilotherm
pdurbin I think you had plans (and an issue) to allow different logins to the same account
14:41 poikilotherm
Maybe that could make this easier...
14:55 pdurbin
yeah, I'm still a little bitter about how that feature got removed
14:58 pdurbin
This is the commit we would need to revert: https://github.com/IQSS/dataverse/commit/fe8e41a
16:04 pdurbin
pameyer: good reminder about security at https://groups.google.com/d/msg/dataverse-community/diQ_Q__tYos/bzSTh8pVBQAJ . Thanks.
16:05 pameyer
pdurbin: no problem; I try to put my paranoia to good use
16:08 pdurbin
just because you're paranoid don't mean they're not after you
16:23 donsizemore joined #dataverse
17:40 pdurbin
pameyer: I hope this make sense and looks ok to you: https://github.com/IQSS/dataverse/issues/5667#issuecomment-474950219
17:41 donsizemore joined #dataverse
17:42 pameyer
pdurbin: it makes sense to me
17:42 pdurbin
good, thanks
17:43 pameyer
it might be helpful to know if that dataset shows up in postgres (with the wrong/new doi)
17:44 pdurbin
yeah
17:44 pdurbin
It got me thinking about files too.
17:45 pdurbin
Your installation handles files very differently than the other ones.
17:45 pameyer
yeah - and I think the dataset that the doi in 5667 resolves to has file bigger than normal dataverse would be happy with
17:46 pdurbin
Storage requirements: 1.6G?
17:47 pdurbin
Oh, you mean her doi? I haven't looked at it.
17:47 pameyer
... and I took another look, and was wrong - 10.7937/K9/TCIA.2015.MI4QDDHU looks to only be a few GB
17:48 pameyer
I was wondering about the user's use of "datacite", and it occured to me I didn't know how to identify which DOI provider registered a DOI
17:49 pameyer
looks like when you use content-negotiation, the redirect target gets set to the appropriate DOI provider
18:04 pdurbin
I'm not sure I'm following you.
18:04 pdurbin
I can swing by if you'd like.
18:05 pdurbin
donsizemore: you're my hero. Thanks for all that Ansible magic: https://github.com/IQSS/dataverse/issues/2073#issuecomment-474901686
18:31 JonathanNeal joined #dataverse
18:54 donsizemore
@pdurbin does that mean i did something right for a change?
18:55 pdurbin
you done good
18:55 pdurbin
and we want more, more!
18:56 pdurbin
Mike loves the "admin1" change, by the way. Thanks for merging it. :)
19:05 donsizemore
<-- lego mindstorm, click "Merge"
19:39 pameyer
top shelf ci pipelines have lego robots in them :)
20:01 pdurbin
pameyer: do you want to reply to our import friend?
20:01 pameyer
you're the expert on the FAKE provider
20:02 pdurbin
heh
20:02 pdurbin
any reason why she shouldn't use her real prefix?
20:02 pdurbin
that's what you do, right? you use your real prefix
20:02 pameyer
right
20:04 pameyer
wait ... just checked, and it looks like that didn't *used* to be necessary
20:05 pdurbin
too late
20:05 pameyer
:(
20:05 pdurbin
I already commented.
20:05 pdurbin
damn my responsiveness
20:06 pameyer
well, it might not have been wrong - I might be mixed up with convigs
20:06 pameyer
s/convigs/configs/g
20:06 pameyer
I managed to hit the size limit in my dataverse ops repo
20:08 pdurbin
dunno
20:10 pameyer
I remember being paranoid about "turn off the working DOI connection", because I had the same concerns
20:12 pdurbin
you only live once
20:15 pameyer
well, once you've manually scripted a song and dance of xml in avnl; it sticks with you as something you'd like to avoid
20:20 pdurbin
:)
20:20 pdurbin
I think I just fixed the d3plus color thing!
20:26 pameyer
nice
20:26 pameyer
.... and 5667 looks to be something else entirely
20:28 pdurbin
the color fix: https://github.com/IQSS/dataverse-metrics/commit/2ee4720
20:28 pdurbin
"heatmap"
20:29 pdurbin
"An Array of colors to use when color values are all positive numbers." https://github.com/alexandersimoes/d3plus/wiki/Visualizations#color
20:34 pdurbin
Any thoughts on if running Dataverse on a "suburi" (new term for me) is possible? Please feel free to comment on https://github.com/IQSS/dataverse-ansible/issues/50
20:35 pameyer
pdurbin: you already covered "don't"
20:35 pdurbin
heh
20:35 pdurbin
asked him to ask the mailing list
20:36 pameyer
5667 is odd. my json works, the json in the issue works, and the requirement I remembered for matching prefixes is definately not a requirement
20:37 pdurbin
odd
20:37 pameyer
yeah.
20:37 pameyer
similarly, the doi in the issue and my dois both work fine
21:02 pameyer
are you using FAKE or DataCite test?
21:03 pameyer
I'm wondering if something changed between 4.9.2 and 4.11 with the create command and datacite
21:47 pdurbin
FAKE
