This is what i am getting:

{“state”:“a4202528-d297-4646-a173-8f89a35abdc8”,“oauth_provider_id”:1,“redirect_uri”:“https://voyager.lemmy.ml/oauth/callback",“prev”:“/”,"expires_at”:1783434115460}

on the response tab i am getting:

error “oauth_login_failed” cause “invalid type: null, expected a string”

You approved the Lemmy login once it redirected to Github right?

yep

aybe try going into your Github settings and removing Voyager entry, then try again

done

And try a different browser, or disabling privacy plugins.

switched from firefox to chrome and disabled extensions, still fails.

Also check the browser console for any errors

I am getting a bunch of them, here is a copy paste:

callback?code=3ed55281d78996fe5c2f&state=82f04f1d-37b2-42fd-b9ae-459081aa151a:9 [unlazy] Missing data-src or data-srcset attribute <img class=​"overflow-hidden pictrs-image object-fit-cover img-icon me-1" src=​"https:​/​/​voyager.lemmy.ml/​api/​v4/​image/​a32c32b6-c8c3-4d17-ab40-46c4ca518868.webp?max_size=96" alt title loading=​"lazy" width=​"96" height=​"96">​ (anonymous) @ callback?code=3ed55281d78996fe5c2f&state=82f04f1d-37b2-42fd-b9ae-459081aa151a:9 (anonymous) @ callback?code=3ed55281d78996fe5c2f&state=82f04f1d-37b2-42fd-b9ae-459081aa151a:9 requestAnimationFrame (anonymous) @ callback?code=3ed55281d78996fe5c2f&state=82f04f1d-37b2-42fd-b9ae-459081aa151a:9 (anonymous) @ callback?code=3ed55281d78996fe5c2f&state=82f04f1d-37b2-42fd-b9ae-459081aa151a:9 9046.client.js:2 Bad element for masonry: .post-listings-grid l @ 9046.client.js:2 i @ 9046.client.js:2 y @ client.js:2 await in y onLoad @ client.js:2 client.js:2 POST https://voyager.lemmy.ml/api/v4/oauth/authenticate 400 (Bad Request) wrapper @ client.js:2 authenticateWithOAuth @ client.js:2 doLogin @ client.js:2 componentDidMount @ client.js:2 (anonymous) @ client.js:2 u @ client.js:2 d @ client.js:2 await in d (anonymous) @ client.js:2 (anonymous) @ client.js:2 3585 @ client.js:2 n @ client.js:2 (anonymous) @ client.js:2 (anonymous) @ client.js:2 client.js:2 Request error while calling POST /oauth/authenticate with {“code”:“3ed55281d78996fe5c2f”,“oauth_provider_id”:1,“redirect_uri”:“https://voyager.lemmy.ml/oauth/callback”} wrapper @ client.js:2 await in wrapper authenticateWithOAuth @ client.js:2 doLogin @ client.js:2 componentDidMount @ client.js:2 (anonymous) @ client.js:2 u @ client.js:2 d @ client.js:2 await in d (anonymous) @ client.js:2 (anonymous) @ client.js:2 3585 @ client.js:2 n @ client.js:2 (anonymous) @ client.js:2 (anonymous) @ client.js:2

I tried to do a sign up using github, still getting a “Login failed.” .

I tried to login using github to voyager and got a “Login failed.” . is it suppose to work?

also it’s getting about 29 users a month. Because lemmy.ml is the official instance of the developers maybe the sidebar on the main page should ask people to sign up and test it? (at least when it is in a period you expect more testing like the beta).

according to the internet he has a return on investment of about 9% in the last 10 years (CAGR). msci world index (which is usually used to compare to see how well a fund performs, and has index funds which track it) made around 13.65% (so a lot higher).

So i would not put too much attention to this guys opinion, sure maybe he is taking less chances (less scary drops in the fund value that some investor don’t like) but looking at the graphs it doesn’t seems that way.

I think it’s a good to do this exercise , a lot of active investing seems like snake oil (although probably not all of it, at least according to some data and research). someone even wrote a book on the subject called “Where are the Customers’ Yachts”.

That is also a product of many variables (including software quality). maybe time to go back to the drawing board.

For this project I’m mainly testing distribution models

What do you expect to learn from those so called tests? (no offence)

The big questions in closed vs open is that there are different scenarios :

for closed source -> less competitor -> more users -> more money -> more investment in the project -> better product.

for open source -> more users want to use it and contribute to it -> better software -> more users -> more potential for making money.

The problem is that for the outcomes you want to track (more money or better software). there are so many variable involved that influence those outcomes so it’s hard to deduce that the license is improving the outcomes or making them worst.

It’s another case of OSPNS - Open source projects names sucking .

List of servers (pretty small)

it’s just that it doesn’t work efficiently enough.

Yeah, but we live in a world of limited resources. in particular labor and specifically knowledgeable linux nerds willing to answers questions for free. If everyone will have that mindset there won’t be a lot of time left to answer the difficult questions .

With that said i agree that occasionally if its done its probably no big deal, there is also linux 4 noobs for those who want to ask some questions to help getting started with linux.

I am considering applying for a grant to fund the development, but there may be complications, especially given that I am a Ukrainian citizen currently living in Ukraine. I am not yet sure what specific difficulties might arise.

Maybe contact framasoft, the creators of peertube. maybe they could help or even provide employment .

Therefore, as an alternative, I would consider fundraising directly from interested users and the community to support development. Since you mentioned fundraising, I would like to ask how you feel about having an option to financially support the project’s development

I recommend giving in a shot, while trying to use lemmy approach of having a good fundraising process with a pop up specifically . I actually tried to write a learning resource for funding open source so there is a lot to say about that. but you would still need a relatively large number of active users. Lemmy for example has around 37K monthly active users and earns around 3.6K a month. so around 10 active users per dollar. I looked at other fediverse platforms and i think it is one of the most effective fundraising processes when measured using active users per revenue ratio.

You could also just put ads in it and maybe a option to pay to remove them . openfront does this and still seems relatively popular. FOSS is nice but eating is better.

Would you personally consider becoming a financial contributor? If so, what monthly amount would you potentially be willing to contribute?

I am afraid i am not a very good case study. i am not really a peertube users and already spend a relatively large chunk of my time on FOSS so i don’t want to invest in it too much.

Regarding your points about benchmarking the recommendation system and curator-based feeds, I am not entirely sure I understood what you meant. Could you please clarify?

Regarding elo ranking something like a page on the website where you are shown two pages with recommendations from two different recommendation systems and you would vote which page and system provided better recommendations . You could also just have one feed and see which recommendation system has a better viewed video to liked video ratio.

Regarding curation. some account invites a bunch of people to act as curators. then something like the most liked videos this group of people had in the last 24h or week get shown. something like that.

If you are looking for free hosting, maybe checkout tuxfamily.

You could also put some links for donations (liberapay , open collective etc) and fundraise the money for better hosting outside of ukraine (and maybe some could go to ukraine government though taxation ). I saw a patreon but no mention that the donations could be used for better hosting which could motivate people.

If you don’t want to deal with taxes, i think using some non profit could help with that (opencollective , Software Freedom Conservancy , spi etc ).

I would like to see some benchmarking of the recommendation system, maybe with a elo ranking like done for AI .

allowing feeds by teams of curators is an idea i would like to have explored (it could be benchmarked using surveys or elo).

This seems really good btw.

They have a explanation here. they claim part of the problem was banning people who are harmless and not homophobes (they show at least one comment ).

Maybe it is better to avoid duplication of effort and contribute to OSGL.

You just need to add a “mature” tag to it and a option to filter by it. contributing to it is easy as far as i can tell.

The website is already linking to google play store and apple store. right now apps that are purely web don’t have a platform to read reviews on . plus neodb lib.reviews are open source although they might not yet be ready for the task yet.

Besides Lemmy mainly gets promoted by word of mouth (eg people recommending it on Reddit)

I doubt that, any data? similarweb shows the top referring site for now is openalternative.co (although at least one of the referring sites mentioned doesn’t seem to make sense for me ).

If people want to review Lemmy communities, it would make more sense to make a Lemmy community for that purpose.

I think people would want to see average ratings. reading a community page means you only read 1-3 reviews and that sample size is too small and potentially biased. you could just run into people who hate a instance for some particular reason (and it’s not hard for me to think of reasons like that).

Two ideas i like:

linking to a bunch of platforms that have reviews of lemmy, fynd does this:

A neodb instance (but maybe another platform would do) where you can read and write reviews about fediverse instances (but that would require at least one volunteer to moderate it).

Using custom libraries sounds like a problem that is easy to miss. sure the super diligent developer will be fine but its like saying there is no point in linters because people just don’t read coding guidelines.

Looks like a few services already have a mechanism like i described in place. e.g. Kubernetes throws a “APIRemovedInNextReleaseInUse”.

We have lemmy apps that still aren’t supporting API changes added over a year ago. We even had one such case last week.

That sounds like something could be improve. is there some sort of warning mechanism in place?

Say when using a lemmy client. the client either specifies its a production build. or if its not then the lemmy server reports where deprecated API’s are used.

Not sure that is the correct approach. break frequently break often seems better (that’s what PHP and java seem to do as far as i can tell, unlike python 3 which caused a lot of drama).

notify a API is deprecated. give some time for users to update to the new API (1 year?) and then remove it.

Of course after version 1.0 there might be less breakage so it won’t be a be problem.