Issue backlog priority order

[alexander0042]

An incredible thank you to @cloneofghosts for keeping things under control while I was celebrating a new addition to my family! Since I'm back now, I wanted to start this note to help keep track of open issues and the order I'm adding them. Short term roadmap is:

Next Version (v2.10.1): September 2026

• Add support for URMA in Time Machine Requests

Version v3.0: Fall 2026

• Add in maps using XPublish;
• Move backend to object storage with caching (maybe on OVHcloud);
• Add in Australia BOM data Source data for Australia? BOM? #361;
• Add HA integration to default repo from HACS
• Review text generation routines to improve response time performance. Goal is processTime < 20 ms.

Version v3.1

1. Marine models (either GFS or ECMWF)

This is my plan, and I'll update here as I work through them! As always, PRs are always open to help fix and address any of these issue, or work on adding new features, especially new data sources).

[cloneofghosts]

You can always add issues to milestones to help track things. I have a few setup but you may need to change some things around as I wasn't sure where exactly things would be addressed. (Added 2.7 and 3.0 milestones so you can start adding things there if you want)

Is this issue still planned for 2.6/2.7 or is something more long-term? #235

I'm assuming this #183 and #361 would be more of a V3.0 update?

I'm assuming this HA issue would be more of a short term fix rather than a long-term one? Pirate-Weather/pirate-weather-ha#358 Same thing with this one? Pirate-Weather/pirate-weather-ha#359

EDIT: Forgot this one as well #387. Likely V3.0? Also this one #195. Would that be 3.0 or included in 2.6/2.7?

[alexander0042]

This is why discussions work well for this sort of thing, since they let me edit the post as I miss things! I added those two HA issues to short term, although the timezone one might be a bit tricky. Added those additional data sources to long term. And the averaged question is probably more likle a v4 thing, since it's contingent on me restructuring all of time machine to optimize how ERA5 data is stored; however, it's a pretty commonly requested feature.

[alexander0042]

Ok, made some changes updates here to align with 2.7.3. Priority is fixing time machine and that midnight bug, which is all working now. That just leaves #472/ #437, which I should get to soon (if I can catch 472 in action). It'll also add some new variables to the ingest scripts, which is necessary to improve the feels like with solar radiation.

My other item is to review @cloneofghosts fantastic Pirate-Weather/pirate-weather-code#163, but at first glace that's looking good and should be able to go into 2.7.3. I really want to unleash AI to just refactor the whole responseLocal.py script, pulling out functions and getting rid of all the "magic number" indexing with a better system, but haven't had a chance to take a look at that yet, Overall timeline is to get 2.7.3 to dev (with these three changes) this week, prod next week.

Last item is the remaining daily issues here: Pirate-Weather/pirate-weather-code#137 (comment)

[cloneofghosts]

@alexander0042 Is 2.9 still scheduled for this month or is it looking more like January at this point? I know Christmas is soon so may not be a good idea to launch anything during the holiday season when you might be away/busy.

[alexander0042]

@cloneofghosts, since 2.9 is more or less live anyway, what do you think of just launching it as a new release, or roll back to 2.8.5? Your summery issues are a really good catch, and I'm seeing a couple other things in the logs, but I don't think there's anything too significant. I have tested the DWD a bit, and I think the new priority there is working, so that's for implementing that.

[cloneofghosts]

I think it would make sense to launch it as a new release.

Maybe see if there's any bugs that can be fixed today and launch it later this afternoon? Unless your busy with holiday stuff today it might be better to launch now?

Two things that I know of that need updating are:

1. The daily precipitation type priority. With mixed being added it still uses the type with highest priority instead of mixed. Similarly sleet needs to account for sleet/ice.
2. DWD MOSMIX still seems to take the average? If you check Rio de Janeiro you'll see what I mean.

[alexander0042]

Ok, two great catches, but both can wait to 2.9.1, and will motivate me to get on it fast (target is in December!). If you can create a new release, I'll push it to prod in the next hour or so and we can go back to normally scheduled operations. I'm looking at the logs from Dev and not seeing any 500's, which is a really good sign

[cloneofghosts]

Sure. I 100% know what this error is coming from Pirate-Weather/pirate-weather-code#484 so I'll take a look soon and merge in a fix.

[cloneofghosts]

@alexander0042 Noticed you've been MIA for almost 2 weeks now but am wondering if the timeline for the next two updates needs to be changed?

[cloneofghosts]

Changed the updates to February and March considering how MIA you've been the past few weeks and am unsure when you will be back.

[alexander0042]

Hey, I'm so sorry for the radio silence! Even by my mediocre standards, this is pretty bad. Goals for the future- post a pinned discussion when I know I'll be away, so I don't leave people hanging. In this case, it was Christmas insanity followed by a busy week packing for a vacation, and then the aforementioned vacation. Back tomorrow though, and then, in an exciting turn of events, I'm off on parental leave for a couple months. Means more time in evenings for this sort of stuff, which I'm genuinely excited about

[cloneofghosts]

No worries. The good news is there was no critical issues with the API while you were gone so that's a plus. Otherwise just minor issues that need to be addressed.

[cloneofghosts]

@alexander0042 For the next major version I was thinking we could add MRMS and the NOAA Enterprise Rain Rate. I started working on herbie templates here: https://github.com/cloneofghosts/Herbie/tree/add-msc-aqi-models (no idea if they work) which could be useful. Was thinking of pysteps for nowcasting and could have MRMS have a higher nowcast frequency then MRMS.

[alexander0042]

Sorry for the radio silence on my end- August is always a busy time, but I had to go to the arctic for some fieldwork and it made it particularly bad this year! Thanks for keeping things moving along while I'm away, and I'll be back at it next week to merge in the Canadian models and take a look at this. I'd love to get some better rain rate stuff in, so that's a perfect place to start.

[cloneofghosts]

No worries! In my opinion the radar stuff should probably be for the next major version and we should focus on finishing up 2.10. Over the weekend I'll finish up the rest of the AQI PR so would just be the Canadian models and the historic PR left.