Time
S
Nick
Message
08:32 jri joined #dataverse
12:32 poikilotherm joined #dataverse
12:50 pdurbin
It feels so good to see actual discussion on the dev mailing list. Like a real open source project. Thanks for replying to this post, poikilotherm - https://groups.google.com/d/msg/dataverse-dev/5hfiXl4Hpd8/awzoSpupAQAJ
12:50 poikilotherm
As promised :-)
12:51 pdurbin
:)
12:52 pdurbin
We just started a new sprint and we're supposed to prioritize the upgrade from Glassfish 4.1, swarming as much as possible.
12:56 poikilotherm
NICE
12:58 pdurbin
Any suggestions on what I should swarm on?
12:59 poikilotherm
Well someone should actually try to find as many errors as possible...
13:00 poikilotherm
To be sure this all runs on Payara, it would be best to really each and every option of what you can do with Dataverse
13:00 poikilotherm
Maybe there should be a test plan?
13:00 poikilotherm
Maybe it would be a good starting point to create UI tests?
13:01 poikilotherm
Because we are in high danger that we break things again while applying a fix for some issue
13:01 pdurbin
I feel like that's been true for years. :)
13:02 pdurbin
We introduce regressions with some frequency, unfortunately.
13:03 pdurbin
Yesterday in backlog grooming we estimated the effort to fix this regression: https://github.com/IQSS/dataverse/issues/6268
13:04 poikilotherm
Oh another idea: you could swarm on reducing the WAR size
13:04 poikilotherm
I'm really eager if we can reduce deploy times that way
13:04 poikilotherm
Which would be very cool for everyone
13:04 poikilotherm
We'll deploy a lot in the next days/weeks ;-)
13:04 pdurbin
I'm all for reducing the war size but I think we should defer that until later.
13:05 poikilotherm
A low hanging fruit in that area would be the Amazon SDK removal
13:05 poikilotherm
And if that is reducing deployment times even by some extent it would be a good indication for later ;-)
13:06 * poikilotherm
goes reading the issue
13:06 pdurbin
yeah
13:06 pdurbin
For UI tests do you prefer Selenium, Cypress, or something else?
13:09 poikilotherm
Whatever works better for you.
13:09 poikilotherm
I just discovered that Cypress is now usable in CI, too
13:09 poikilotherm
They released Docker support
13:11 poikilotherm
Ah Jenkins pipeline example :-D https://github.com/cypress-io/cypress-example-kitchensink/blob/master/Jenkinsfile
13:11 pdurbin
We've experimented with both.
13:12 pdurbin
- selenium: https://github.com/IQSS/dataverse/blob/v4.18.1/tests/test_create_test_account.py
13:12 pdurbin
- cypress: https://github.com/IQSS/dataverse/blob/v4.18.1/tests/cypress/integration/test_login.spec.js
13:12 pdurbin
no opinion? no preference?
13:12 poikilotherm
When going with Selenium, you should be using JUnit + Selenium ;-)
13:12 poikilotherm
Cypress looked like the new cool kid on the block
13:13 poikilotherm
https://docs.cypress.io/guides/references/trade-offs.html#Permanent-trade-offs-1
13:17 pdurbin
"The sweet spot of Cypress is to be used as a tool to test your own application as you build it. It is built for developers and QA engineers, not manual testers or exploratory testing."
13:21 poikilotherm
https://blog.logrocket.com/cypress-io-the-selenium-killer/
13:21 poikilotherm
Sounds like it would be much quicker to go with Cypress
13:21 poikilotherm
Easier to maintain and use
13:21 poikilotherm
No cross browser support, but does that really matter?
13:22 poikilotherm
And the specs are a lot more easier to write with avoiding all those pesky waits and sleeps
13:25 poikilotherm
So please go for Cypress.io ;-)
13:25 poikilotherm
My biggest fear was about not having it in CI, but this is all fixed
13:37 pdurbin
What about Slava? He keeps talking about Selenium. Would you be able to talk with him and get his thoughts on Cypress vs. Selenium?
13:39 poikilotherm
You could reach out to him by mail ;-)
13:39 poikilotherm
https://github.com/IQSS/dataverse-sample-data/issues/13#issuecomment-565010526
13:45 poikilotherm
pdurbin: what do you think about nightly builds... Should I only use the tags develop and develop-payara? Should I add a date and keep some configurable amount of builds so one can easily switch?
13:53 pdurbin
Those tags sound fine. And sure, nightly sounds good.
13:53 poikilotherm
I looked around to find some other projects doing nightlys
13:54 poikilotherm
Looks like many provide a separate image to distinguish from stable releases
13:54 poikilotherm
I found no example of adding dates to tags...
13:54 poikilotherm
Seems like no one wants to maintain retention policies ;-)
13:59 pdurbin
What's the phrase? Fall forward? When you push a bad deployment and rather than rolling back you push another deployment to fix your regression.
13:59 poikilotherm
:-)
14:06 pdurbin
Someone in #crimsonfu probably knows. I'm asking there: http://irclogs.shortcircuit.net.au/%23crimsonfu/2019-12-12.log.html
14:11 poikilotherm
pdurbin: about that password thing: we talked about I would hand over :-D
14:13 poikilotherm
Thx for adding the hook
14:46 pdurbin
sure, just pointing out that it hasn't happened yet
14:47 poikilotherm
Tell me how, to whom and when ;-)
14:50 pdurbin
It should probably be a GitHub issue.
15:38 pdurbin
poikilotherm: I'm asking Slava about Cypress in Skype
15:38 * poikilotherm
thumbs up
15:39 pdurbin
"Oliver prefers Cypress" --me
15:41 poikilotherm
?
15:42 pdurbin
Am I misrepresenting you? :)
15:43 poikilotherm
Nope
15:43 poikilotherm
I was confused by that "--me"
15:44 poikilotherm
Sounded like an option to an unknown command :-D
15:44 pdurbin
s/me/pdurbin/
15:45 pdurbin
poikilotherm: he's writing Selenium tests in a private bitbucket repo.
15:45 poikilotherm
Ok
15:45 poikilotherm
Gotta run guys...
15:45 poikilotherm
Kids waiting
16:26 stefankasberger joined #dataverse
23:10 iqlogbot joined #dataverse
23:10 Topic for #dataverse is now Dataverse is open source research data repository software: http://dataverse.org | IRC Logs: http://irclog.iq.harvard.edu/dataverse/today | Who's who: https://docs.google.com/spreadsheets/d/16h3jv24usMGq18495C-JA-yNcQCKiKDa65MTraNDd7k/edit?usp=sharing
