#amara

maggie_s: did you activate a demo on link demo?

all: back

maggie_s wb

all: wrapping up, later folks!

ehazlett: I tried to, but it made link.demo 502 as well, so I deactivated the link

arthurdebert: later! have a good one

maggie_s: which demo?

ehazlett: prod-test

maggie_s: fixed

ehazlett: awesome, thanks a lot!

ehazlett: is the link activated right now?

maggie_s: yes

yw

ehazlett: terrific! will it be safe to change it later if another env requires linking?

maggie_s: i hope so -- ipython was installed which caused the link to fail -- i'm sending the dev team an email stating to not install that until i can fix the bug

ehazlett: ah, I see :)

all: hey

ehazlett: yeallow

ehazlett morning

arthurdebert: hey

all: morning

sjl: how do?

all: morning

nicksergeant: how do?

nicksergeant morning

arthurdebert: can I get some more DMR tickets?

bendk: I'll assign you a few if you want

bendk sure, before you do, would' you mind looking at the jenkins build failure?

bendk the diffing change broke a bunch of tests

bendk sjl it would be nice to have both dev and staging branches passing tests (see janeted '

s email

yeah

where sure

how can I log into jenkins?

bendk it's open if you just want to look at the results http://dev.universalsubtitles.org:8888/

bendk I believe you can create an account though, let me know if you can't

deanjansen hey, I'm doing work on the billing stuff for fastrack, maybe you can clear things up for me

deanjansen we had an 'old' system (what's on prod), and we're implementing a new one (2124)

arthurdebert: yep

arthurdebert: we were going to have them both up in parallel

so we can compare results

deanjansen ah, dear god

deanjansen ok, hm, got it

arthurdebert: and once we're confident in new system, ditch the old

arthurdebert: is that difficult? I thought they were completely separate… it's what craig and I had discussed

deanjansen wouldn't it be better to QA it well then keep two parallels bunch of code?

deanjansen not difficult, everywhere in the code there are two almost identicals way to do things, just confusing,but definitely not a problem, will do

arthurdebert: the reason to test in parallel is then you're testing on production data during a production billing cycle

if anything seems majorly off, we should easily be able to spot it

deanjansen I see

arthurdebert: and having Terri doing the QA in this instance is pretty valuable

deanjansen got it

deanjansen for the second billing system, we should have a web UI that generates it as wll right?

arthurdebert: yeah, I don't know if that's been done at all yet

deanjansen it hasn't at all

sjl - can I ask you a few questions regarding tasks and delete - I'm working on tests and trying to unravel the mysterious book of i2005 - repost - I dropped off

2005 - DMR: Connect UI to 'visibility override' instead of 'delete' - https://unisubs.sifterapp.com/issue/2005

janeted: oh lord... sure

lol

i know - i'm sorry

heh, it's not your fault

actually probably won't be so bad

visibility is the trickiest part of the new data model (aside from the tasks stuff that's grafted on as a holdover from the last data model)

so if I set up my versions like in your explanation - v1 - private, v2 - private, v3 - public, v4 - public

yep

Then I unpublish v4 - with Delete optoin

Then I upload a de translation

should it's lineage point to v3 or v4?

lineage should point to 4

ok - but if I were to open the translation dialog - I would see the text for v3?

as should the parents

yes

well

maybe

nooooo

if you don't, that's a bug

jaja

ok

I'm not 100% sure it's working correctly now

but yes, what you describe is what it *should* do

is there a way to see what the text would be, without have to hover over the stupid task and actually open the translation dialog?

some field or value that will help me?

the text that would be in the dialog?

yes

or something that I could tell what version is being used.

um

Not that I know of

I guess I'm not understanding why it's lineage would be v4 if it had been deleted. Is v4 the new v3?

ok, so let's back up to some ideas about the new data model

in theory I can do the hover and open the dialog - it's just that those stupid things are technically hidden in css and it's a total pain in the ass

the first commandment of the new data model is that aside from visibility and visibility_override, SubtitleVersions are immutable. they must never change except for those two fields.

ie - selenium doesn't want to deal with hidden ui elements

ok - thou shalt not delete.

no editing the subs, no deleting a version, no modifying anything about it except those two fields

restricting them like that makes things *much* easier to work with in a LOT of places

so

if you "delete" v4, it's still going to exist in the database

now, lineage

lineage is something we came up with to solve the following problem/question:

Project unisubs build #5335: STILL FAILING in 10 min: http://dev.universalsubtitles.org:8888/job/unis...

Steve Losh: Ignore more stuff.

"I'm looking at German version 2. What has changed with the languages that people have used in the past to make these subtitles?"

E.g. if someone used English to help them make german in the past, and English has changed, then we should tell them the next time they edit the German subs, so they can look at the changes and see if they need to fix anything.

Doing this just with the parent foreignkeys would be possible, but would perform poorly because you'd need to look at the parents of every version of German in the DB

"Was German v4 based on anything? No, okay, how about v3? v2? v1? Oh yes v1 was based on english, ok let's check if English has changed."

that make sense?

yes - I appreciate the fact that you are a good clear writer

hah, thanks. I'm wordy but it's usually better than my terse mode

Okay, so we need a way to tell if something's changed since the last time this version was created. That's where lineage comes in

When you use a version to help you make another version, that version's code/number gets added to the lineage map.

And this map gets added to over time.

So, in our example, when you create German based on v4 of english, it will be German v1, parents english v4, lineage {'en': 4}