#concourse

:D

dubkat_: btw you can download fly with that fix from https://ci.concourse.ci/api/v1/cli?arch=amd64&a...; now

@xoebus do you have an idea as to why that unmarshal fails?

I wanted to ship that

I was also wondering who else uses atc.Duration cc @alexsuraci

zachgersh: sorry, no idea

seems like no one is using atc.Duration

Why can't I just change that field to interface (so it displays nicely)

I think it would keep it as 10s as your interval showed in your example

sorry, meant 1m in your example

I think because it's a time.Duration behind the schenes

*scenes

yeah atc.Duration is just a var for time.Duration

but I don't see any other fields on the config structs using it - so why is it even a thing?

and if you want it to display like '1m' wouldn't it be better to just be interface or string and then do a configvalidation?

yeah possibly

it'd be nice to have the validation at configure time though

you guys already have a package for that right?

it unmarshalled it completely and pulled it into a duration

before applying it

mapstructure might be doing something

it's an atc.Duration so that we know how to unmarshal it

otherwise it only permits parsing an int

in nanoseconds, which sucks

Looks like a few more stories before the 0.57 release marker. Is there any hope we'll get the new garden release before the weekend? (Preferably with a day to test it out. :p)

We're hoping so

The garden team have been super fast at fixing our issues

So we may get that out on Thursday

I need to check on the progress of the other track

It'll be tight with the shorter week

Cool.

We're using the bosh deployment resource to test out our stacks against cf-release. We'd like the resource to upload --rebase. Would this not be desired, or perhaps it could be an option? We don't really want to have to track the cf-release version numbers and bump them when we make our stack modifications.

I guess our version doesn't have to match cf-release. Maybe?

mhoran: we won't be adding --rebase; it's a pretty janky bosh feature that last i heard they're planning to remove once there are better options. which i think there are now.

the version can be whatever you want (that's what --rebase does; it creates its own version based on the current latest)

one trick is to just version it as a unix timestamp

Okay. Cool.

i.e. create release --version 0.0.$(date +"%s") or whatever it is

mhoran: are you testing against non-final cf releases?

Yeah; acceptance-passed.

Timestamp sounds reasonable.

We use timestamps elsewhere. I was definitely over thinking this one

in an ideal world there would be final releases coming out of runtime's pipeline much more frequently

having everyone create their own cf-release is a nightmarish hellscape

Our use case is a bit weird. We're taking over rootfs and need to test that against the release. So we have to create a release anyway.

fair enough

We ran out of files again.

@alexsuraci http://i.imgur.com/MFbzauk.png

finally, figured out the new location and grouping business to allow better rendering of these. The thick borders are, ok, probably need to be tweaked

Cool. What're the thick borders? (And what happened to the padding on the left of the aggregate?)

the thick borders are taking the place of the padding, it could stiill be a 1px border with 9px of padding too

Yea I'd prefer just the 1px border added to the left, and shifting the step to the left by 1px so its still aligned.

Which would apply to all steps; no special casing for aggregates

I thought you wanted aggregates in hooks rendered differently

Nope, same as everywhere else

The border is enough

same as in, 10px of padding (except for the left) and the grey

the left being, 1px of border and 9px of padding

Same padding and grey, with the border, but unindented by 1px so its all still aligned (as in margin-left -1px)

So the border sticks out to the left

all still aligned as in , all the hook borders line up

with 10px of padding

for the aggregate group

also, an interesting effect has entered into the mix with re-working the origin: http://i.imgur.com/ZodhAC0.png

nested hooks

Yeah that's the effect I'm going for

Let's you read long task output and know the full stack of context just by looking at the borders to the left

By still aligned I mean hooks should align with things like the get in put-get. And not be pushed over by the border

yup

But yea the borders should align too

think I got ya, hopefully, until I push and it all goes wrong

Lol. Sweet

Yusss

shipit

ha ha, still has a bit of work on it, and like, tests for the new event data being passed through

but the hard parts I think are done

cool