imax (@imax.ooo)

imax @imax.ooo
[ View ] Jul 11, 2024 at 08:11 UTC

0 replies 0 reposts 1 likes

imax @imax.ooo
[ View ] Jul 10, 2024 at 20:43 UTC

Yeah, I get that, and would immediately do it. However, it's the same problem that I have with vscode: I *kinda* can turn off annoying features/bugs that get in my way, but I'd rather have a decent text editor out-of-the-box than this

0 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 10, 2024 at 20:41 UTC

That would annoy the fuck out of me 😅 That's why my smartwatch doesn't pretend to be *that* kind of smart, and doesn't do shit I didn't ask it to

1 replies 0 reposts 1 likes

imax @imax.ooo
[ View ] Jul 10, 2024 at 19:21 UTC

*sigh* I hope they don't fuck it up too badly

1 replies 0 reposts 2 likes

imax @imax.ooo
[ View ] Jul 10, 2024 at 19:19 UTC

github.com/bluesky-soci...

0 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 10, 2024 at 18:57 UTC

Protocol design 101: If there's a natural way to make invalid configurations impossible to express - you take it

0 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 10, 2024 at 18:54 UTC

tl;dr: facet offset can point to a middle of a UTF-8-encoded codepoint

0 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 10, 2024 at 18:52 UTC

Because this way graceful degradation doesn't insult reader's eyes with a bunch of arcane markup. I do, however, take offense at *how* they decided to count the offsets

1 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 10, 2024 at 18:50 UTC

FYI, since recently you can unpin "Following" and move BlackSky to be the first tab

1 replies 0 reposts 1 likes

imax @imax.ooo
[ View ] Jul 10, 2024 at 15:25 UTC

Fascinating and sobering reading: www.ic3.gov/Media/News/2...

0 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 23:48 UTC

Although, in the current implementation these lookups are designed around too, and happen in the background every so often, outside of request serving path

0 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 23:45 UTC

It would quickly become a SPOF and a major source of request latency, if not designed around.

1 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 23:38 UTC

Also, handle -> DID lookup history is not recorded anywhere, so PLC history alone is not that reliable in this regard

1 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 23:34 UTC

PLC does keep history, but I don't think anything actually uses it in a meaningful way. Facets do indeed record the DID, but the record content is immutable, so the old handle is still shown, but clicking on it leads to the correct profile

1 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 23:30 UTC

ugh, yes. You've already got a password input field provided by your platform, don't mess with it

0 replies 0 reposts 1 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 23:29 UTC

I've ended up treating handle as a display name. Even then, if you use your own domain name as a handle - it's still portable between storage hosts. (Don't recall exact details on portability of `*.bsky.social` handles, but that can work just fine too)

1 replies 0 reposts 1 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 23:16 UTC

Paypal does this and I don't have any problems using it with Bitwarden. I just trigger the browser extension twice. (Speaking of Paypal, why the fuck does anyone think that allowing only one U2F token is a good idea?)

0 replies 0 reposts 1 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 23:09 UTC

I'm almost afraid to ask, but what steps does one need to implement to remove the "support" for password managers?

1 replies 0 reposts 1 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 23:06 UTC

Did the journalist ask you what does it taste like?

0 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 23:01 UTC

Nah, I could get behind a Steve, but Tim is just some wannabe, fuck him

1 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 22:58 UTC

- How deeply are we fucked? - Yes.

1 replies 0 reposts 4 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 22:55 UTC

um, what

0 replies 0 reposts 2 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 20:19 UTC

Did I mention already than Chrome is the new Internet Explorer?

0 replies 0 reposts 1 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 20:09 UTC

TBH, this world would be a better place is people were growing pigs on their farms instead of writing code

0 replies 0 reposts 2 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 18:20 UTC

You reminded me of the time I ported Prometheus metric pusher to OpenComputers, so I could monitor my energy and items storage 😅

github.com/imax9000/Ope...

1 replies 0 reposts 1 likes

imax @imax.ooo
[ View ] Jul 09, 2024 at 08:59 UTC

Got tired of running `zfs list` to see where did the disk space go and wrote a short script to feed it into `ncdu`

github.com/imax9000/zfs...

0 replies 0 reposts 1 likes

imax @imax.ooo
[ View ] Jul 08, 2024 at 20:34 UTC

i.e., if you want to count them as missing - just treat 500 as 404. If you want to additionally check that it's not just the relay being dead - poke `/xrpc/_health` after you get 500

0 replies 0 reposts 1 likes

imax @imax.ooo
[ View ] Jul 08, 2024 at 20:20 UTC

500 roughly means "yes, there is or was such repo, but now it's unavailable" I'm not exactly sure which way you want to count them in your specific use case

1 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 08, 2024 at 20:18 UTC

That seems consistent with what I'm saying: > 404 means it's unknown, 200 or 5xx - known

1 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 08, 2024 at 19:47 UTC

For Go there's a thin wrapper that falls back to plc.directory if your mirror is not caught up or returns an error for some reason

github.com/uabluerail/i...

0 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 08, 2024 at 19:44 UTC

Now you can just use it as a direct substitute of plc.directory in your DID resolver

1 replies 0 reposts 1 likes

imax @imax.ooo
[ View ] Jul 08, 2024 at 19:43 UTC

BTW, indexer includes a proper plc mirror:

github.com/uabluerail/i...

disable all containers other than `postgres` and `plc`, expose port 8080 on the plc container, start it, give it couple hours to download the full history - and done

1 replies 0 reposts 1 likes

imax @imax.ooo
[ View ] Jul 08, 2024 at 19:33 UTC

1 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 08, 2024 at 19:31 UTC

For unknown DIDs?

1 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 08, 2024 at 19:30 UTC

Actually getLatestCommit has very similar logic around this, so you can keep using that one

sourcegraph.com/github.com/b...

1 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 08, 2024 at 19:28 UTC

You can do getRepo (and skip reading the full response). 404 means it's unknown, 200 or 5xx - known

sourcegraph.com/github.com/b...

1 replies 0 reposts 1 likes

imax @imax.ooo
[ View ] Jul 08, 2024 at 19:17 UTC

("because PLC rate limits" is a valid answer. But you can also just run a local mirror)

0 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 08, 2024 at 19:15 UTC

Why not resolve the DID and ask the hosting PDS directly?

2 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 08, 2024 at 18:53 UTC

(making it optional assumes that actually setting it has a semantic value, e.g., branching off a new thread. But we don't have that, so best would be to not have it there in the first place)

0 replies 0 reposts 0 likes

imax @imax.ooo
[ View ] Jul 08, 2024 at 18:50 UTC

Correct protocol-level solution to this would be to make "root" optional. It is redundant with "reply". Having both is an optimization that should've been done entirely inside the AppView, but it also creates ambiguity in what a client should do when they're set inconsistently with each other

1 replies 0 reposts 1 likes