Skip to content

Retrospective‐Notes‐2025.04.02

Jack Harper edited this page Feb 4, 2025 · 2 revisions
Chair Timekeeper Note Taker
TW JH

Present:

  • In person: IG GR KB
  • Online: JH TW ES CMS SC DK FA LJ LC JD

Previous sprint

nothing - short sprint

Current Sprint

PI changeovers can be confusing

Christmas PI changeover was tricky - if there are any ideas to make this easier please suggest.

shared group laptop

it's very helpful (in fact it may even be another "excellent helper"?) to take to meetings and to the halls

ticket priority

medium and low have been picked up ahead of high priority tickets this sprint. In this case the release tickets were dependent on each other, but that doesn't explain all of them. we should pick from the highest priority downwards regardless of if we want to do it or not.

putting code in the source code of the dev manual

we should move these out to the code repository, not the wiki repository, as they are a bit of a pain to see otherwise as you have to clone the wiki repository.

manual system tests on release

are the statuses currently in place enough? should we split into fail, fail - fix, fail - noted? Yes. Issue proposed to do this and improve the template: https://github.com/ISISComputingGroup/IBEX/issues/8633

build nodes

they are slow. we should axe off NDW1757, it's terrible. Can we have faster build nodes? we could build over the network and split things client/server as servers have caches. The main bottleneck for build nodes at the moment is disk speed, perhaps we need more node. we could consider using dev machines overnight as jenkins nodes. Would have to be careful about soak testing etc. RIP NDW1757

feedback from scientists - confusing "instrument patching" / "ibex deployment" email

can we combine these better so that its less confusing? we had the opposite request last time. we don't normally send out "done" emails as it's just extra admin. perhaps we should start doing so?

galil-old is bad

yes, it's bad, we should test the new one, it's hard to maintain and keep features up to date. galil-old caused issues over the last release as it relies on VS2010. the new galil driver isn't quite there yet, we need to test on CRISP (they're happy for us to do so)

email sent to unclear

it was an old FIT email - we'll delete it.

nice that we can pip install genie_python now

yes

first line support

we should remind people who are on first line support the next week - we'll start doing this at standup.

zipping up builds

sophos is horrible and tries to read the millions of files that eg. client builds are made up of - we should zip these up. we should also do this with the ibex backup from NDX to data-old. turns out we already do this! but we should make our nightly builds not bother with leaving the unzipped build. that will save a lot of files

twincat on developer machines

we should not bother installing it on developer machines, and leave NDXMOTION to run twincat XAR so it can simulate beckhoffs.

Mad/Glad/Sad

glad - release done!

glad - pat testing done!

glad - interviews done!

glad - we won't get timing system issues this cycle (allegedly)

mad - the release was not fun

mad - partial derivatives and asymmetry is hard

Clone this wiki locally