Time
S
Nick
Message
08:08 poikilotherm joined #dataverse
08:16 juancorr joined #dataverse
08:32 MrK joined #dataverse
10:57 pdurbin joined #dataverse
11:26 pdurbin
Hi, everyone.
11:28 pdurbin
poikilotherm: I see at https://github.com/IQSS/dataverse/pull/5680#issuecomment-475868350 that you don't like the idea of moving Swift settings from a file to JVM options. You like the status quo? Carlos trying to improve performance.
11:28 poikilotherm
Hi pdurbin :-)
11:30 MrK
Hi
11:32 pdurbin
You should all come to LibrePlanet next year. It was fun. :)
11:32 poikilotherm
:-)
11:33 poikilotherm
About my comment: you can use the domain.xml approach, it should be ok with ConfigAPI
11:33 poikilotherm
But you really should avoid "-" and "_"
11:33 poikilotherm
Use dots :-D
11:33 poikilotherm
And camel / pascal case
11:34 pdurbin
like our other JVM options
11:34 poikilotherm
Otherwise it will be VERY difficult or at least tedious to generate settings from scripts, etc.
11:34 poikilotherm
Yeah
11:34 poikilotherm
Currently, there are two options with dashes
11:35 poikilotherm
I need to create an issue (&PR?) to change that
11:35 poikilotherm
(Old names should be aliases of course)
11:37 poikilotherm
I just updated my comment
11:41 pdurbin
Do the other JVM options have to do with Swift?
11:41 poikilotherm
Nope
11:41 poikilotherm
dataverse.auth.password-reset-timeout-in-minutes
11:41 poikilotherm
dataverse.files.hide-schema-dot-org-download-urls
11:44 pdurbin
Ok. So nothing to do with Swift. I'm just thinking about the QA effort. If it can all be combined. Yes, please create a separate issue, if you would be so kind.
11:44 poikilotherm
Sure. That's why I created the issue in dataverse-kubernetes first, so this doesn't get lost
11:45 poikilotherm
pdurbin: I'm setting up the ORCID sandbox...
11:45 poikilotherm
Interesting comments in the code:
11:45 poikilotherm
// It's a design decision to not suggest a username.
11:45 poikilotherm
public String getNeverSuggestUsername() {
11:46 poikilotherm
return null;
11:46 poikilotherm
}
11:46 pdurbin
Huh.
11:46 poikilotherm
I am not sure this will find friends here :-(
11:49 pdurbin
Please see "Make a final decision branch on suggesting a username or not..." at https://github.com/IQSS/dataverse/issues/3338
11:49 pdurbin
"Make a final decision branch on suggesting a username or not for OAuth users (ORCID, Google, GitHub). When you log in with an institution (Shibboleth, HarvardKey) you are automatically assigned a username under the idea of "don't make me think." We could make all "remote auth" options (ORCID, Google, GitHub, institution/Shibboleth/HarvardKey) consistent when we work on #3486. When you create a
11:49 pdurbin
local/builtin account we do not suggest a username based on what they type for first and last name. Deferring this and will revisit all places in the app where usernames can be suggested."
11:56 poikilotherm
Ok.
11:57 pdurbin
:)
11:57 poikilotherm
Did I miss sth. in the docs about groups? I just found groups from IPs and from Shibboleth
11:59 pdurbin
Nope. Groups are not well documented: https://github.com/IQSS/dataverse/issues/2412
12:02 poikilotherm
I see.
12:04 poikilotherm
There is an API for those, right?
12:05 poikilotherm
Because it might be interesting to write a script to get groups from LDAP/... and automate stuff
12:06 poikilotherm
Ah, got it in the docs
12:08 pdurbin
great
12:10 poikilotherm
Groups are inherited from a parent dataverse to a sub dataverse, right?
12:12 poikilotherm
Oh, they are not. Ok...
12:24 poikilotherm
LOL I think I found an interesting inconsistency
12:24 poikilotherm
You cannot see an unpublished sub dataverse in the root dataverse even when you are an admin of that dataverse
12:25 poikilotherm
Even though the root dataverse has "Member" role for the user
12:25 poikilotherm
Accessing the dataverse via a link is successfull
12:28 pdurbin
It sounds like we don't index the permissions into Solr properly.
12:31 poikilotherm
Sounds like a bug
12:31 poikilotherm
Yeeeahaaa
12:31 pdurbin
:)
12:31 poikilotherm
I'll create an issue
12:31 pdurbin
thanks
12:53 poikilotherm
Any ideas for searching if this has already been reported?
12:54 poikilotherm
https://github.com/IQSS/dataverse/issues/3471
12:55 poikilotherm
That sounds familiar...
12:56 pdurbin
heh, you beat me to it
12:56 pdurbin
can you reproduce that issue?
12:58 poikilotherm
Actually I more or less did that, but with another role
12:58 poikilotherm
Will post a comment to the issue
12:59 donsizemore joined #dataverse
13:04 pdurbin
sounds good, thanks
13:05 donsizemore
@pdurbin morning. will touch up PR, but I thought you wanted the curl examples modified in a separate PR?
13:06 poikilotherm
pdurbin: https://github.com/IQSS/dataverse/issues/3471#issuecomment-476187458
13:06 pdurbin
donsizemore: morning. Oh. Well, I was saying (or trying to say) that my style is to slowly, slowly fix the examples over time. I do the same with removing tabs from the Java and putting spacs in instead. :) Please do what you feel! :)
13:07 pdurbin
poikilotherm: thanks, added to https://trello.com/b/YI9EbqGz/backlog-grooming-for-dataverse-pdurbin-edition ( donsizemore, this is the new https://github.com/pdurbin/druthers :) )
13:07 donsizemore
@pdurbin i'll do each in separate commits in the same PR. jenga?
13:08 pdurbin
donsizemore: sorry, I'm confused. For this pull request, please just fix up the curl commands that are on topic (destroy or whatever) for what QA needs to test.
13:08 pdurbin
Does that make sense? I don't want to greatly increase the QA effort.
13:08 donsizemore
👍
13:09 pdurbin
phew, thanks :)
13:11 pdurbin
donsizemore: also, just to re-re-confirm, sample data should be in code review, right? Or are you still making commits?
13:11 donsizemore
@pdurbin i thought it was ready for code review, but that was a couple weeks ago. kick the tires, then kick me if you need
13:13 pdurbin
thanks, dragged it over
13:15 pdurbin
I had a nice time hanging out with dzho yesterday by the way. It turns out he's in a lot of IRC channels. We got in a nice debate about science vs. open science. :)
13:30 poikilotherm
This does not sound like its working ok....
13:30 poikilotherm
[#|2019-03-25T13:29:07.185+0000|INFO|glassfish 4.1|edu.harvard.iq.dataverse.search.IndexServiceBean|_ThreadID=26;_ThreadName=http-listener-1(1);_TimeMillis=1553520547185;_LevelValue=800;|
13:30 poikilotherm
unable to index dataverse. id was null (alias: test2)
13:30 poikilotherm
Happened while creating a new dataverse
14:53 pdurbin
Any other details? I just created a new dataverse as of 3c836a980 (develop branch) and I didn't see that indexing error.
14:57 pdurbin
donsizemore: still there?
14:59 donsizemore
@pdurbin yes sir
15:00 pdurbin
I just approved your pull request but there are merge conflicts. Do you mind if I resolve them?
15:00 donsizemore
@pdurbin oh, poop. you're welcome to but i thought i updated the fork again this morning
15:01 pdurbin
I don't think your branch has the new "Dataset Metrics" stuff from Make Data Count.
15:04 pdurbin
ah, maybe I'll tweak your formatting a bit while I'm in here.
15:06 pdurbin
and switch to -H "X-Dataverse-key:$API_TOKEN"
15:10 pdurbin
ok, done, pushed
15:10 pdurbin
and moved to QA
15:23 andrewSC joined #dataverse
16:04 pameyer joined #dataverse
16:37 pameyer
pdurbin: `scripts/deploy/phoenix.dataverse.org/` is still the place to check for what phoenix is doing for deployments, right?
16:44 pdurbin
yep
16:45 pdurbin
I mean, there's a little logic in Jenkins too.
16:45 pdurbin
A test that a published file can be downloaded.
16:48 pameyer
but if you were doing server side stuff for the sushi example log, you'd be doing it in that directory? or somewhere hidden in jenkins configs?
16:49 pdurbin
Oh, duh. Sorry let me think a minute. I'm debugging my Python. :)
16:50 pdurbin
I said duh because my first thought was "does this work over the wire?"
16:50 pdurbin
But is there a wire for you?
16:51 pdurbin
Does the file get copied to /tmp ok?
16:51 pameyer
mvn on the host system, glassfish in the guest/container
16:51 pdurbin
I'm not sure if I'm helping.
16:51 pameyer
yup - host has it in /tmp/
16:51 pdurbin
hmm
16:52 pameyer
although oddly with a modification date of 3/18.....
16:52 pdurbin
odd
16:52 pameyer
once I'm seeing if I've got the favicon in the right place, I'll try deleting it and making sure it gets put back
16:53 pdurbin
ok
16:56 pdurbin
wow, 13 errors now? I just saw your comment: https://github.com/IQSS/dataverse/issues/5662#issuecomment-476282992
16:57 pameyer
yeah; but all because of one file
17:02 pameyer
and I'm guessing nobody ran the test suite between when that got merged and this morning...
17:02 pdurbin
I don't know when it got merged.
17:02 pdurbin
4 days ago
17:02 pdurbin
probably not
17:03 pdurbin
I don't know. I didn't.
17:04 pameyer
... inferring from context that flyway got merged before mdc too.
17:05 pdurbin
I don't remember, sorry.
17:05 pameyer
not important
17:06 pameyer
as usual, might be off base - but it would be consistent if jenkins/phoenix showed the same mdc error
17:07 pameyer
but flyway means that's not something that can be checked w\out postgres upgrade on that box
17:07 pdurbin
yeah, all servers need to be upgraded to postgres 9.3 at least, 9.6 recommended
17:08 pameyer
server.log does mention that it doesn't have /tmp/sushi_sample_logs.json
17:08 pameyer
... and putting it back has things passing
17:11 pdurbin
hmm
17:11 pdurbin
does the copy not work because src/test stuff isn't available?
17:11 pameyer
right - because `src` isn't on the server
17:12 pdurbin
ok
17:12 pdurbin
where should we put it?
17:12 pameyer
;) I was just going to ask you
17:12 pdurbin
:)
17:12 pdurbin
resources?
17:12 pameyer
I'm not sure if there's a test utility for "push this to glassfish from local filesystem" - but I'm guessing that file needs to be on filesystem server-side
17:14 pameyer
so I'm not sure if it makes more sense to fix in the deploy scripts or fix it in code
17:15 pdurbin
I'm not sure either. sushi_sample_logs.json is loaded in MakeDataCountUtilTest but there are no assertions. So it's not really testing much. Unlike the test in MakeDataCountApiIT which has a lot of assertions.
17:24 pameyer
it probably doesn't belong in `src/main/webapp/resources` - that would get a test data file deployed to production
17:24 pameyer
I don't know if `src/test/resources` would get server-side or not though
17:25 pdurbin
src/main/webapp/resources/lang/languages.json
17:25 pdurbin
src/main/webapp/resources/dev/sample-shib-identities.json
17:25 pdurbin
do those show up in production?
17:27 pameyer
sure looks that way to me
17:27 pameyer
I'd missed src/main/webapp/resources/dev/
17:27 pameyer
you woudn't see them in production unless you went looking for them, and it's unlikely they'd break anything
17:28 pameyer
seems kinda hack-ish though
17:48 pdurbin
yeah
17:48 pdurbin
where is the json for creating datasets? used by the api test suite, I mean
17:49 pdurbin
scripts/search/tests/data/dataset-finch1.json
17:49 pdurbin
hmm
17:49 pdurbin
pameyer: why would you have "scripts" but not "src/test" in docker-aio?
17:50 pameyer
pdurbin: I'd guess a mix of trying to keep the build context small, and duct tape
17:51 pdurbin
heh
17:51 pdurbin
Do we think phoenix will have "src/test"?
17:51 pameyer
I think I only have `scripts` in there because of a lack of understanding of what the test setup actually needed, and figured throwing the kitchen sink in was acceptable
17:52 pameyer
my impression is that phoenix doesn't have `src/test` - but I don't actually know
17:53 pdurbin
Ok, your gut feeling is probably right.
17:54 pameyer
I took a quick look - mdc was merged ~ a day before flyway was. which is consistent with the idea that this would've showed up there
17:54 pdurbin
I don't know if this helps but one of my Jenkins jobs scp's the "scripts" directory (and the war file) from the Jenkins server (where the war file is built) to the phoenix server.
18:28 pdurbin
donsizemore: I'm realizing I don't like "http://$SERVER" because it encourages http instead of https.
19:05 donsizemore
@pdurbin i'm looking at your tweaks and realizing i need a better markdown editor (which are you using?)
19:21 pdurbin
vi
19:21 pdurbin
but rst isn't md
19:21 pdurbin
do you have `make html` working?
19:22 pdurbin
doc/sphinx-guides/Makefile
19:28 pameyer
resist the temptation to try and use `make html` and a html validator to find rst errors you made. I've made that mistake at least 3 times
19:50 donsizemore joined #dataverse
20:43 pdurbin
:)
21:26 poikilotherm joined #dataverse
21:38 pameyer
@poikilotherm since you mentioned orcid sandbox, did you see the mention of the cert fiddling in the guides for orcid?
21:39 poikilotherm
Ola!
21:40 poikilotherm
Yarp
21:40 poikilotherm
At least sandbox had no problems
21:40 poikilotherm
Without any fiddling
21:40 pameyer
... I can't distinguish either of those as yes or no
21:40 poikilotherm
Ola was spanish for hello ;-)
21:40 pameyer
but if logins worked with no problems, your glassfish ca trust certs are new enough
21:41 poikilotherm
I used plain Glassfish 4.1 install
21:41 pameyer
gotcha ;) I learned to spell it "hola"
21:41 poikilotherm
Oh that was a typo
21:41 pameyer
hmm....
21:42 poikilotherm
Or portuguese (but then a typo because of the missing Ž)
21:42 poikilotherm
I will keep this in mind for later
21:42 poikilotherm
Thanks for pointing it out :-)
21:43 pameyer
no problem. I'm glad it worked
21:43 poikilotherm
Me too
21:43 pameyer
heading out; but I'm puzzeled that it worked for you without updating the trust store
21:43 poikilotherm
Cu
