Haiku, R1/Beta3 - Test Candidate 2 availability

I’ll spin up a usb stick and give tc3 a road trst

Wait, who suggested that ?! We just said that we could release with some known bugs. I mean, in beta2 we had an easy to trigger app_server crash when using the web browser. This bug is much less bad.

I didn’t. I can’t even reproduce this bug at all on my machine. Everything works perfectly here. So it happens only sometimes, on some machines, and on some websites. And a bit of scrolling up/down eventually fixes it.

So, if someone has an idea how to fix this, well, please submit your patch and we can go ahead and integrate it. But for me it is not a critical bug that should delay a release for months and months, which is the trajectory we are on currently, because no one has any idea what this bug really is.

Also, WebKit can be updated easily separately from the OS, so when a fix is found, it can be deployed to all users.

So it’s all nice everyone agrees it should be fixed. Do you have an idea about the problem? What if no one does? What if it takes us two years to figure it out? After pushing us for months to ship a release, suddenly this one single bug (out of hundreds of others) becomes super important?

It’s a beta release. That means it ships with known bugs. It’s work in progress. That’s normal. Otherwise we will never put any release out.

9 Likes

One, potentially quick hack-fix could be to trigger a scroll by one pixel up and down, without delay between, right after the page has been loaded. Or maybe a “redraw” of the browser window’s displayed content. I have no knowledge about what the process is to actually show the rendered content but my I idea is simply to “refresh” that state (not reload the url ofc) after page load.
I have no idea if that’s really feasible from your point of view. However, I’d submit a patch but I don’t know my way around (I looked to get an idea on if I could do it) the source. So if that’s a reasonable approach someone would need to work on that patch. Just an idea.

No, the bug is too unpredictible. Sometimes it will appear on the first rendering, but sometimes it will appear when scrolling down. Or not at all.

The first thing would be to understand exactly when this state is triggered, but once we understand that, I think the problem will be a lot easier to fix

1 Like

Yeah, of course that makes sense. And for an actual fix the cause must be understood.
I understand that adding quick-hacks is not a good approach as that creates tech-dept and stuff becomes even harder to maintain.

Earlier in this thread the suggestion/possibility was brought up to you by @brunobastardi

It was a question because of the “?” question mark!
What is your answer useful either? The question was answered by @PulkoMandy already. You don’t have to answer this question. It is only rhetorical one!

Why thank you, for pointing the obvious out so efficently

I’ve just installed the TC2 build and noticed that the Brazilian Portuguese language strings aren’t up to date (some of the outdated strings were changed three weeks ago). Shouldn’t all translation data be synced in this build?

The latest RC1 image has updated translations. This is an older test build of R1/Beta3. With that said, i’m going to lock these older threads to prevent confusion. Feel free to resume chatter in the latest R1/beta3 RC1 thread :slight_smile:

2 Likes

See R1/Beta3 RC1 for the latest stable test build.