Recent stability issues

[gyzerok]

Hello there!

Let me start but saying that by no means this is a rant. I have a great respect for the core team and I love working with svelte. I've also got my first contribution approved and was super happy to finally contribute back to the framework!

Recently (like maybe a couple month) it feels like stability is somewhat declining. Some issues are getting fixed and some new ones appear. There are currently 6 issues with p0 label some of which are dating back to the end of April.

We are using svelte heavily in production and currently are in a bit of a weird spot. We can't really downgrade because we are using some newer features like attachments and need some newer bugfixes, but we also can't really upgrade to get some of the issues fixed because something new gets broken. Also it gets increasingly difficult to hunt for reproductions. It was never easy, at some point I've spent a week to find what is causing a memory leak, but now it feels like it's even harder.

We've started using svelte about a year ago and back then it was sort of the opposite. It felt like every new release was providing greater stability and performance.

Maybe the core team is also noticing this? Are there any opinions on the matter?

Back in the days it was also extremely helpful when we had some calls with Dominic and he was helping me figure out reproductions for some memory leaks, so he could fix them. I am wondering if something like this could be possible now with other maintainers?

Thank you!

[dummdidumm]

Hey, sorry for the increased amount in bugs/regressions lately - I'm as mad as you that we've lately introduced some new bugs by fixing others. We are committed to get these gnarly reactivity issues fixed.

Which issues specifically hurt you / block you from upgrading? For which do you think jumping on a call would be helpful? Feel free to ping me on Discord!

[gyzerok]

Thank you Simon, really appreciate you jumping in on this!

About the current issues

This one is currently number one error for us in crash collection system. Meaning not connected to Svelte, but out of everything. And in every case when I look at the code it should be impossible due to TypeScript checks.

Another one is connected to unowned signals. I was even asking about it during the Svelte Summit panel, but I think my question was misunderstood. The immediate problem we've had, I fixed by wrapping things into $effect.root to make them owned (never figured repro for our specific case). So there are no known problems in our codebase regarding this, but what makes me worried is if there are unknown or there will appear new, cause code is getting written. Also we are using classes and global state which sort of steers us towards unowned signals more.

The third one is the one where I can't provide even a reproduction. The only thing I know is the version from which it starts breaking and suspect this PR to be the cause. Now awaiting some pending work to try and upgrade us again to see if maybe it was connected to something else that got fixed.

Also I just discovered that we are running half legacy mode even though we don't intend to. I try to follow issues and PRs quite closely to keep us updated and it never occurred to me that issues and fixes related to legacy mode could affect us. So now my intention is to force rune mode to reduce our bug surface area. I've tried it, but our code breaks due to state_unsafe_mutation. Fortunately I was able to contribute half of the fix myself and Paolo is of great help figuring out the second half!

This last one is what I am waiting right now to attempt upgrade us to see if maybe the third one is gone.

About your call offer

Mostly I think that would be helpful in cases where I can reproduce something in our codebase, but have zero success making a repro out of it. At least for me it works such that I can often guess what the source of the problem is by the description of a problem from my colleagues. So here I sort of think that even though there is no minimal repro you could make an educated guess based off your knowledge about internals. Definitely it was the case with Dominic and we had great success with couple of tricky bugs back in the day by just discussing it over the call and trying things.

Currently it seems like there is nothing like this, the question was more about if that's something that is available. Appreciate you offer and will contact you if it feels like that could help!

[gyzerok]

Also through our year of Svelte usage I noticed that you seem to rely on tests a lot. And the suite is definitely great. However sometimes new versions are getting released and they break stuff, but then the next version fixes other bugs. And we end up not being able to upgrade to get the later fixes.

The most difficult part about it is when when can't get a repro for the newly introduced bug and it's not like you can revert releases. In this cases it becomes a bit stressful. Not sure if anything could be done about that. Like maybe some RC process? But then again it'll slow things down. Anyway just wanted to share these thoughts.