mastodon.gamedev.place is one of the many independent Mastodon servers you can use to participate in the fediverse.
Mastodon server focused on game development and related topics.

Server stats:

5.1K
active users

#dostodon

3 posts2 participants0 posts today
Replied in thread

@root42 I _could_ make releases of #DOStodon, but just uploading a working directory with the Javascript sources and the #DOjS binaries is way simpler.
After all ist is a side project that spawned from a bad joke!

And it took me weeks to port #DOjS to win32, which I consider my time well spent instead of adding releases to #DOStodon...

Oh noes, I was deducted "cool points" and now my open source project (which I maintain all by myself w/o any help or contributions in my free time) is less cool because a self proclaimed GitHub repository inspector does not like how I release my software!!!one!eleven

What shall I do? Should I delete #DOStodon? Should I delete myself? The world?

I like how they have made the effort to tell me they won't use it instead of providing a PR or help to fix it...

Replied in thread

@kim

I think someone got #DOStodon running on a 386 with 9MiB.
They used a TLS proxy though, because the 386 is to slow to do the SSL handshake and the server closed the connection.

You can try a 486 with 66MHz and 8MiB RAM if you are adventurous and have lots of time.

Anything Pentium and 16MiB should work, but is quite slow...

@FediTips

Nachdem #DOStodon ja wirklich gut auf meinem alten Thinkpad R60 laeuft habe ich angefangen mal etwas tiefer in den Kaninchenbau rund um #dojs abzutauchen...
Javascript unter DOS ist irgendwie eine faszinierende Sache, eigentlich die logische Fortsetzung von BASIC oder anderen interpretierten Sprachen von "damals"

I’m doing some preliminary investigation into the #Macstodon login issue on Mastodon 4.3. I haven't found the specific cause of the problem, but the error message (“Security verification failed, are you blocking cookies?") seems to imply a CSRF failure.

I'm not entirely sure why the vintage Mac web browsers I've tried are unable to satisfy this requirement - maybe WebOne is mangling the CSRF cookie, I have to do more testing here.

One approach to solving this problem would be to use the "password” OAuth grant type, rather than the "authorization_code” grant type that is currently being used. Basically, you would log in to Macstodon with your Mastodon username/password directly, instead of using a web browser. This is the same approach that is currently being used by #DOStodon (github.com/SuperIlu/DOStodon/b) and some other retro clients.

However, there are two problems with using the "password” grant type:

1. Not all servers support it. For example, compare the outputs of mastodon.social/.well-known/oa and urusai.social/.well-known/oaut - and note that mastodon.social lists "password" under "grant_types_supported”, but urusai.social does not (also, that endpoint doesn't exist prior to Mastodon 4.3, so if the server is running an older version you just have to perform trial and error to see if password grants are supported)

2. MFA is not supported. If you're using multi-factor authentication with your Mastodon account, your only option is to log in using the "authorization_code" grant type. There is no option for appending your MFA token to your password.

The other option is to use a hybrid approach, where Macstodon would continue to use the "authorization_code" grant type, but would prompt for a username and password as if it were using the "password" grant type. Rather than launching a web browser, it would make the request for the login page directly, fill in the user's credentials and post the form, then screen-scrape the result for the authentication code. This is the approach currently being taken by #MastAppleII (github.com/colinleroy/a2tools/).

This approach is more brittle, and a much more challenging development effort, but it should work with most servers as well as for users with MFA.

Of course, this approach has one big gotcha of its own - older and newer Mastodon versions are less likely to work correctly due to changes in the design of the login page. Perhaps the "password" grant type should still be offered as an option (if the server supports it), and we allow the user to choose how they want to log in? Or would that be too confusing for users? Alternatively we could try the hybrid approach first, and then fall back to the password grant type if that fails.

Anyway, I'd love to hear your thoughts on how to approach this!

RE: oldbytes.space/@smallsco/11331

GitHubDOStodon/mstdn.js at main · SuperIlu/DOStodonMS-DOS Mastodon client. Contribute to SuperIlu/DOStodon development by creating an account on GitHub.