#20 consider another binary name a month ago

Comment by ~mvdan on ~eliasnaur/gio

which is very close to the go tool with a slant towards GUI programs.

I'm confused. I assume plain 'go build' for Android produces a binary, while I assume that the current gio produces an APK.

To continue the bikeshedding: goport, gopack, gopherapp.

#20 consider another binary name a month ago

Comment by ~mvdan on ~eliasnaur/gio

Note that the same name clashes (such as Roger's issue) happen with 'go install' too, so I don't think this issue is about Linux distro packaging per se.

What's special about your build tool, compared to 'GOOS=android GOARCH=arm64 go build'? Perhaps that's going to help us come to a reasonable name.

#20 consider another binary name a month ago

Comment by ~mvdan on ~eliasnaur/gio

I'm afraid gg is far too common, and already taken in the space of Go tools: https://myitcv.io/cmd/gg

I think ggo is a bit too close to 'go', and could confuse some people.

ggio could work. Almost like forcing people to pronounce the tool in one way :)

#28 end to end testing a month ago

Ticket created by ~mvdan on ~eliasnaur/gio

I think this project should have proper end-to-end tests to be more stable and maintainable in the long term. That is, automate the checking that the wayland/windows/etc ports actually work properly.

Perhaps I could help by writing the first end-to-end webassembly test. I maintain chromedp, which can be used to drive Chrome via pure Go, so I think that would be a good option for the use case. The test would simply be skipped if Chrome isn't installed.

If that seems like a good idea, I can send a small patch. Ideas are welcome on what kind of features we could test as a start; I was thinking simply an app that changes background color when clicked anywhere. That would be pretty trivial to test via chromedp, issuing a click at a fixed coordinate, and getting the color at the same coordinate.

#1 apps: replace the gophers demo a month ago

Comment by ~mvdan on ~eliasnaur/gio

May I suggest something like what you showed at GopherCon UK? For example, a text input and button to add custom labels to a list from top to bottom. This can show:

  • How to use a text input
  • How to use a button
  • How to dynamically add elements to the UI
  • How to lay out elements from top to bottom

I think this covers the basics, and would go further than the existing hello trivial example.`

Bonus points if it scrolls too, but I wonder if that would make the example too complex.

#27 Using 'go run' with a different target a month ago

Ticket created by ~mvdan on ~eliasnaur/gio

Suppose I'm on Linux with X11. At the moment, if I 'go run' an example, it errors, as X11 isn't supported and it tries and fails to connect to a wayland server.

If I start Sway under X11, 'go run' suddenly starts opening windows inside Wayland, which makes sense.

However, once X11 is supported, it would be great to be able to tell 'go run' what available target to use. I imagine that any of these should be possible on my Linux setup above:

  • Native X11
  • Wayland, e.g. under X11
  • WebAssembly, via $BROWSER

Or is this out of the scope of 'go run'? If so, how does it pick what target is appropriate?

#126 patch emails without git's format signature are confusing a month ago

Comment by ~mvdan on ~sircmpwn/lists.sr.ht

Also, please realise that while I'm used to both email and git, I'm not used to combining both. This issue might seem obvious to developers who are used to git format-patch and send-email, but evidently it wasn't obvious to me :)

#126 patch emails without git's format signature are confusing a month ago

Ticket created by ~mvdan on ~sircmpwn/lists.sr.ht

I failed to send a patch over email for over two hours, because lists.sr.ht kept showing my email in the archives tab, but not the patches tab.

It took me a while to realise that the issue was this tiny part of my user's gitconfig:

[format]
       signature =

Granted, I don't even remember why I set this in the first place years ago. But I think sourcehut could make little mistakes/errors like these more obvious to the user.

For example, if an email has [PATCH] in its subject but fails to parse or be recognised as a valid git patch, perhaps lists.sr.ht could show an error message or warning of some kind. Without that pointer, I was investigating my SMTP server for over an hour, guessing that the issue had to do with some setting like the content encoding.

If this is out of scope for what sourcehut should be doing, then my alternative suggestion is for git-send-email.io to mention to users what gitconfig entries could cause issues with email patches, such as format.signature.

#26 OpenGL Widget a month ago

Comment by ~mvdan on ~eliasnaur/gio

In the long term, wouldn't simply supporting Vulkan be very portable? Even on OSX these days, one could use MoltenVK to get it working there, and Windows/Linux/Android machines generally ship with Vulkan support today.

#24 ui/app: two worrying staticcheck warnings a month ago

Comment by ~mvdan on ~eliasnaur/gio

Formatting take 2:

./ui/app/internal/gpu/gpu.go:307:6: this value of data is never used (SA4006)
./ui/text/label.go:58:3: this value of x is never used (SA4006)