Comment by ~mvdan on ~eliasnaur/gio
I agree that the current method isn't perfect, but at the same time, the tests are end-to-end. Running a binary to perform the build should not be a problem, nor should it noticeably affect the cost of the tests.
If you do this for the sake of simplifying dependencies and the tests, I think it might be worth it. If you do this to speed up the tests, I don't think you'll get compelling results. Many of the tests run expensive programs that make an exec call comparatively trivial.
Ticket created by ~mvdan on ~sircmpwn/go-bare
$ git grep -n 'bare:' cmd/gen/main.go:164: return fmt.Sprintf("`bare:\"%s\"`", name) example/schema.go:21: Name string `bare:"name"` example/schema.go:22: Email string `bare:"email"` example/schema.go:23: Address Address `bare:"address"` example/schema.go:25: OrderId int64 `bare:"orderId"` example/schema.go:26: Quantity int32 `bare:"quantity"` example/schema.go:27: } `bare:"orders"` example/schema.go:28: Metadata map[string][]byte `bare:"metadata"` example/schema.go:40: Name string `bare:"name"` example/schema.go:41: Email string `bare:"email"` example/schema.go:42: Address Address `bare:"address"` example/schema.go:43: Department Department `bare:"department"` example/schema.go:44: HireDate Time `bare:"hireDate"` example/schema.go:45: PublicKey *PublicKey `bare:"publicKey"` example/schema.go:46: Metadata map[string][]byte `bare:"metadata"` example/schema.go:68: Address [4]string `bare:"address"` example/schema.go:69: City string `bare:"city"` example/schema.go:70: State string `bare:"state"` example/schema.go:71: Country string `bare:"country"`It would be useful if the main README file explained what these field struct tags are for. From reading the spec, I assume they're not used for the encoding and decoding. I skimmed the codebase and they only seem to be used for code generation.
Comment by ~mvdan on ~sircmpwn/scdoc
I was about to file this issue. It seems like ronn renders backticks just like bold asterisks, so we could do the same.
Ticket created by ~mvdan on ~eliasnaur/gio
For example, see https://www.recursive.design/, which is a font that's highly customizable. It allows modifying all of these variables in a scale:
- monospace (from 100% sans to 100% mono)
- linear vs casual
- weight (light vs extra-black)
- slant (regular vs cursive)
This is particularly interesting for Gio, because if one wants to ship an entirely self-contained app, it's a good idea to ship fonts as well. But if one wants a monospace font for code, a sans font for English, and a cartoon-ish font for something else, you likely need to ship three fonts.
With a variable font, you could ship a single one and then tweak the parameters in each of the UI components. Note that I have very little experience with font rendering, so I don't know if this would be a big refactor, or simply exposing this font as a package.
HN discussion: https://news.ycombinator.com/item?id=23929151
Comment by ~mvdan on ~sircmpwn/hub.sr.ht
I agree with this issue; the current UI isn't great for the majority of projects, where there's just one git repo and one issue tracker.
A redirect isn't our only option, though. The page could also simply list some of the issues, such as the most recent ones, under the link to the issue tracker. This could also be done for any number of issue trackers; you could think of it as an "activity feed" of issues only.
on ~eliasnaur/gio
REPORTED RESOLVED FIXED
Comment by ~mvdan on ~eliasnaur/gio
I did this in e97adeedd9186530b84a40ebd0ebb7af66eb0e07.
~eliasnaur I thought the "fixes" comment lines would work in sourcehut automagically, so I guess they don't.
Comment by ~mvdan on ~eliasnaur/gio
real Xorg + wine + gio: works real Xorg + wine + dxvk + gio: works virtual Xorg + wine + gio: works virtual Xorg + wine + dxvk + gio: does NOT work
Comment by ~mvdan on ~eliasnaur/gio
We figured out the cause:
real Xorg + wine + gio: works real Xorg + wine + dxvk + gio: works virtual Xorg + wine + gio: works virtual Xorg + wine + dxvk + gio: does NOT work
It appears that Wine's built-in Direct3D implementation works on virtual Xorg servers without DRI3, but DXVK won't work properly when DRI3 isn't available. Xephyr and Xvfb don't seem to support DRI3.
We'll fix this by making the end-to-end tests use a completely separate WINEPREFIX, so that the user's config (such as DXVK being installed) doesn't affect whether or not the tests pass.
Comment by ~mvdan on ~eliasnaur/gio
I should clarify that by "it fails to render properly" I mean that the gio app refuses to start or show a window. I'll use this thread to post more information in the future, if I have any.