I have had a go with creating resources to fill the cards and I think it’s safe to say that it’s unlikely to evolve into anything worthwhile so I likely won’t be adding anything into the cards. I don’t think that it’s too uncommon though, take for example cards seen on OpenAI’s website:
And the Twill CMS page:
It might make more sense though in the contributing section to just opt for 3 cards side by side that are a lot smaller and function similar to the cards seen in the OpenAI example
Not necessarily, no. An automated build step and easy updates would be nice. The main issue I can think of is that apps.ankiweb.net/downloads hosts a mirror of downloads for people who can’t access GitHub, so the downloads and landing page would need to be split into different domains.
I’m glad static deployment is supported, but this is not a complicated website - what value is Astro bringing to the table that justifies the extra moving parts?
Each extra format/technology we add to the mix is a little extra complexity, which adds up over time. It’s not necessarily a dealbreaker in this case, as the .json files are simple and Pontoon appears to support them, but Fluent would be preferable.
If we’re going to be handling internationalisation whilst still serving a majority of static content, Astro seems to me like the best fit given it’s design principles and features. I don’t really know of any other viable approaches.
I’m unsure how we would integrate the Fluent translations though. I can do some research but initial findings show it’s not well documented, especially for use in Astro. Something like Fluent feels a little overkill though for simple sentence translations that won’t have dynamic content.
If we want to keep things simple, I think SvelteKit SSG is also an option.
Could you please provide some concrete examples, so I can better understand your argument? All we need to do is load the appropriate language file, and get the appropriate translated strings so we can {interpolate} them. That can be done quite easily with a raw Svelte or SvelteKit. What is Astro doing that they can’t?
If your argument is just “I don’t know Fluent”, you can just stub out the translation-fetching function for now, and one of us can add it later.
I don’t have any experience using fluent, though I have looked into the unofficial svelte package that can handle the localisation. I wasn’t entirely certain how the internationalisation would work out for static content with SvelteKit as surely the translations rely on dynamic interpolation of the string based on the user’s locale. All of which relies on checking conditions client side. Either way I don’t fully understand the workings of this so if this is a non-issue then that’s fine.
Astro is purpose built for this use case, on the other hand SvelteKit is designed for creating SPAs with dynamic content. Either way I don’t particularly care what we use, as long as everyone’s on board and it gets the job done. Both projects have good DX and support static content generation, so if SvelteKit is preferred, we can proceed with SvelteKit.
SvelteKit is used by AnkiWeb already, and also with the current desktop beta, so using it for the landing page as well would mean we avoid adding an extra technology.
Sorry for the slower updates on my behalf, I’ve had a lack of spare time recently. I’m hoping to create the hero section placeholder graphic / image soon-ish. I have redesigned the testimonials section to be a carousel, that way any number of comments from the community could be showcased, no matter their brevity. I also took inspiration from the previously shown OpenAI cards in the new contributing cards.
I created the ‘wireframe’ graphic for the decks section in the getting started section. I’m quite happy with how it looks and intend to do something similar for the reviews section. Would be great to hear some feedback / thoughts before I get started on that though.
Hey @dae as someone mentioned it last year, the URL is confusing too. You can either end up in the landing page or in ankiweb depending on whether you added a “apps.” in the front. Can something be done about it too? I also want Anki to be more popular in my country. Having a concise URL thus would be good.
Looking good!
You can tell people ankisrs.net if you wish. I’m not really keen to move the website to that domain, as it requires a separate SSL certificate, and SRS doesn’t mean anything to many people.
I am of the opinion that the landing page should be on ankiweb.net and ankiweb itself (that we use) should be on apps.ankiweb.net
What do you think?
Sorry, I disagree.
Recently I’ve been trying to decide on a suitable image/graphic for the hero section. I had a go at trying to recreate a graph showcasing the forgetting curve and how spaced repetition affects retention, but it’s certainly neither proportional or accurate so I don’t wish to proceed with it:
I was trying to roughly imitate the graph as seen on An Annotated Forgetting Curve | Teach Like a Champion :
I thought this was likely the most fitting graph to showcase the power of spaced repetition which is why I was intending on featuring it in the hero. As for alternatives, I considered using the RMSE graphs that showcase how FSRS v4.5 outperforms other algorithms, though I don’t think it’s suitable for all audiences.
Does anyone have any suggestions on what might be an effective hero section image/graphic? In the meantime I will try to work on my illustration skills…
IIRC FSRS wiki had a forgetting curve graph. Can you check?
I don’t see the problem with showcasing FSRS. The messaging should not be that Anki is outperforming other apps when it comes to scheduling, but rather that Anki includes SOTA spaced-repetition-scheduler, FSRS 4.5, and then you include how accurate it is at predicting retention* and that it can reduce workload by upto 50%! Let’s see what Dae says.
*This should be rephrased as most people won’t understand the significance of it but I don’t know how we should go about it.
Thanks for the tip, I believe the graphs you are referring to are on these pages: Spaced Repetition Algorithm: A Three‐Day Journey from Novice to Expert · open-spaced-repetition/fsrs4anki Wiki · GitHub
My only concern with using others’ content is the licensing / permission and ideally the illustration would need to match the colour scheme in use
I totally agree but the problem as I see it, is that you can’t convey this through just the graph alone. Having knowledge of what RMSE and it’s significance is would be a requirement to understanding the graph, which I fear non-technical audiences would fail to have.
However, maybe we could also consider adding another short section to the design to showcase some of these figures. The reduction in workload is a great point to make as this resonates with all audiences. Maybe we should decide on ~3 of these statistics to feature prominently?
It’s from osmosis dang. But this does serve as a template. After a few edits it’s indistinguishable from others right? unless these come with DRM.
I think a link to the benchmark repo itself is better: GitHub - open-spaced-repetition/srs-benchmark: A benchmark for spaced repetition schedulers/algorithms
People who care about technical details can read all the technical stuff. People who don’t care or aren’t math-savvy won’t benefit from seeing a bunch of arcane numbers.
I don’t know man. I’m thinking about how companies market their smartphones for example. “Look at our new phone! It has an octa-core processor!” why would they do it if it doesnt attract non-techie customers? (this is real example)
