[Dart/Flutter]: v9

[buenaflor]

Description

Topics:

• Dart version bump to 3.5.0, therefore Flutter 3.24.0
• Cleanup wasm compatibility and internal API
• After dart bump we can safely bump our Sentry Android SDK dependency to v8

Success Criteria

• Hard to define proper success criteria but stability should be the main goal -> check SDK crash detection
• Get full pub dev points back - we lost some because of the WASM hack

Post-GA tasks:

• Add replay as an option to flutter wizard sentry-wizard#826

Update on 11 April:

• All main tasks are completed
• Currently we are looking into things that we might have missed
• Beta will follow

Update on 22 April:

• There have been delays due to uncertainties of the exceptiong group feature and issues with the release tool as the auth token changed
• Updated due date to May 7

Update on 24 April:

• Beta released
• Preparing changelog and migration docs

Update on 6 May:

• Second beta was released
• Added fix for changing handled to unhandled for onError
• All in all a bit delayed due to the fix above
• Will begin an RC this week

[Rexios80]

This would be the perfect time to migrate from hive to hive_ce in sentry_hive

Please reconsider #2400

I see you are working on WASM compatibility in this version. The original hive does not support WASM, whereas hive_ce does. The hive_ce package also has other convenience features over the original hive and is about to get isolate support with IO-Design-Team/hive_ce#69, which is a highly requested feature (see felangel/bloc#2633, isar/hive#1285, isar/hive#1167, etc).

[buenaflor]

@Rexios80 hey thanks for bringing this up, I agree this is a good opportunity. I'll bring this up internally

[buenaflor]

We discussed this internally and we will keep support for hive as well as for isar. Reason being the adoption of the community edition is still not large enough to warrant a switch. hive: 1.22m downloads vs hive_ce: 90k on pub.dev, we will definitely keep an eye on it as the gap closes. time will tell what will happen but for now we'd like to align with the stats and support as many devs as possible. and supporting a 2nd package with only ce is also out of scope for us currently.

What we will do is add additional checks and tests for compatibility with the community version so users can feel safer in using it (if they decide to use it). Additionally we can also add some custom code snippets so users can copy paste code that instruments it with Sentry if they are not comfortable in overriding the original hive dep.

[Rexios80]

I don't know what the stats have to do with it. The second users realize that Hive doesn't support modern Flutter use-cases such as WASM and isolates, they will want to switch. Hive has 5 years on Hive CE and was extremely popular, so of course it's going to have "better" stats.

If you really want to look at stats, you can clearly see the like graph for Hive plateauing:
https://pubstats.dev/packages/hive

While the graph for Hive CE is doing the opposite:
https://pubstats.dev/packages/hive_ce

(look at the year/all time ranges)

Although I do appreciate the extra care you plan to give to users of Hive CE, I see this as a misstep in the maintenance of your package. If you are postponing migration in the hopes that the original Hive will rise from the dead... it won't. Even if Hive v4 ever does get released, that would also be a breaking change (probably requiring a new package) for you since data generated by Hive v2 will not be compatible with Hive v4.

Hive v2 data is fully compatible with Hive CE.

[buenaflor]

It's not about hoping whether or not hive rises from the dead, I don't have any personal biases here, we are taking the stats as fact that right now hive is more widely adopted (based on weekly downloads which averages > 1m) and if at some point the opposite is true or is nearing the gap then of course we will revisit this.

If hive_ce's growth really explodes the next months then for sure even publishing another package (sentry_hive_ce) is possible if there is a huge demand, it's not like we are blocked here until our next major and I personally would prefer that anyway so we can support both sides.

The arguments for transitioning make a lot of sense I agree and don't get me wrong it'd be nice to migrate but it's still early - it's a tough situation, ideally we'd like to accomodate everyone but for now we will stick to supporting the majority and do we what we can on the side to help out ce users