[GSoC 2024] Porting WebKit2 Final Report | Haiku Project

Hi Zarshard!
I’m new to Haiku and just wanted to see the implementation of mouse support, but the link that you have provided is not redirecting to the actual page.

Hello harsh!

That’s weird that that link no longer works. Perhaps it’s because the haiku-webkit2 branch is rebased every time we pull in code from upstream? Still, I would have thought that kind of URL would last forever. Oh well :man_shrugging:

I’m pretty sure I was referring to this commit. It’s on the latest rebase but might not be in a month!

For posterity, if that link breaks, the name of the commit is “Implement mouse support” and it modifies, among others, the Source/WebKit/Shared/NativeWebMouseEvent.h file. If you can’t find it, feel free to ask me again :slightly_smiling_face:

It would if there’s still something pointing to it somewhere on github (a tag, a branch in a fork, …), but maybe we forgot to set up a tag at the end of the project? And the webkit2 brancheindeed is regularly rebased and slightly modified, so it will not point to the old commits anymore.

Makes sense. Yeah, I don’t recall us ever making a tag to it. I should see if it’s on my computer so that I can reupload it.

So, there is a HaikuWebKit2-GSoC2024 tag. It looks like all of my links in the final report still work, including the link to the commit implementing mouse support that we’ve been talking about. It’s being kept alive, I believe, by PulkoMandy’s tag as well as my merged pull request, so it should be good for at least a decade (fingers crossed).

Fortunately, I still have that commit that no longer exists on GitHub on my local machine. Now I’ll need to decide whether and how to reupload it (the hard part is figuring out what the latest descendent of that commit is so that I upload as much as possible of that branch instead of just that commit and its ancestors).

Link fixed! I uploaded the commit and the children that I could find. Ironically, even though the tag that is keeping the link alive is in my repository, it still means that that commit now exists in the parent repository as well. At least GitHub warns that the commit “may belong to a fork outside of the repository.”

1 Like

Yes, that’s an effect of how github manages forks. They made that a very cheap operation (since it’s at the core of how github works) and so the repository isn’t actually duplicated, it’s just different set of branches and tags pointing to the same repository

Is there Mini Browser – the WebKit2 version, I mean – in such state to be distributable / installable via pkgman … for some probing ?

1 Like

No, you have to compile it yourself if you want to try it.

There is not much interesting to probe anyways. Many websites crashes including google home page. Rendering is flickering. Mouse wheel scrolling doesn’t work. The UI is extremely simple and doesn’t even work right (fails with no error if you forget to put http:// in the url bar, typing enter in the url bar does not load the website, instead you have to click the “Go” button, etc). This is, of course, in addition to the existing problems with WebPositive, as there is no reason any of those get fixed magically by just using a different API to access the same web engine.

At this point, user testing will not be so helpful. Give us a little bit of time to get something set up that is actually useful and working somewhat correctly :slight_smile:

4 Likes

I see, right. Thanks for your reply. :sun_behind_small_cloud: