@jenniferplusplus@hachyderm.io
@jenniferplusplus@hachyderm.io avatar

jenniferplusplus

@jenniferplusplus@hachyderm.io

trans lesbian
staff software engineer
devops, reliability, resilience, sociotechnical systems
also, that gay shit

Was https://tech.lgbt/@jenniferplusplus

This profile is from a federated server and may be incomplete. Browse more on the original instance.

thisismissem, to fediverse
@thisismissem@hachyderm.io avatar

Is there any software that allows instances to be at different pathnames, instead of different hostname / port combinations?

Like:

https://social.example/foo => one instance

https://social.example/bar => a totally different instance

As far as I know no, but I'm looking at the spec, and I'm not sure there's anything in there about this?

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@thisismissem there's a lot of assumptions that the domain provides a sort of namespace. Like, it would be pretty hard to share the webfinger path, or nodeinfo.

danilo, to random
@danilo@hachyderm.io avatar

BeefstickAI, which wants to build generative AI for dogs and doggy daycare providers, raised a $14M series A led by Gullible Ventures and LPSwindler (TechCrunch)

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@danilo I see no evidence those (hypothetical, for legal reasons) guys can or did actually generalize their realization to all women. Their behavior is better explained by trying to protect their property (daughters, and more to the point, their future grandsons)

hrefna, to random
@hrefna@hachyderm.io avatar

One idea I've toyed with before that I think is really interesting is the idea of having a persistent inbox but an offline processor

So think of it like this:

You have a queue that sits online with a webserver in front. The queue has some logic (the kind you can configure with RabbitMQ) for deduping and routing, but is still just a queue.

That's the only persistently online piece.

(@jenniferplusplus mentioned this earlier as well, this is not unique nor original to me as a thought)

1/

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@hrefna I've specifically imagined this as a 3rd party service.

But yeah, I also have notions of doing this internally to handle excessive demand spikes, ironically because of the lack of backpressure mechanisms.

jenniferplusplus, to random
@jenniferplusplus@hachyderm.io avatar

I'm making sure that Letterbook runs both as a self contained single binary and as a horizontally distributed system for high availability and capacity. This is going to be worth it, but at the same time I've created an incredibly obnoxious packaging problem for myself and I'm really sick of it.

Like, really sick of it.

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

Basically, the plan is to support any of these configurations. Letterbook is designed to scale both up and down, in a highly cost conscious way.

In terms of software development, this isn't even hard. What makes it hard is packaging. And what makes packaging hard(er) is that most people don't even conceive of it as something to be done.

Box and arrow diagram of multiple instances of a simple service, behind a load balancer. Modules are the same as the simple deployment. There is an added RabbitMQ message broker, and the Bus module passes messages to that, which are then sent to the Worker modules. SSL termination is handled by the load balancer. It anticipates serving 10s of thousands of users and devices. Labeled as a simple load-balanced deployment of Letterbook
Box and arrow diagram of a moderately complex distributed system. The API, Web, and Worker modules are broken out as isolated services, and scaled independently. The load balancer distributes requests to the Web and API services. The message broker distributes messages only to the Worker services. SSL termination is handled by the load balancer. It anticipates serving a million users and devices. Labelled as a distributed, scaled, high-availability deployment of Letterbook

jenniferplusplus, to random
@jenniferplusplus@hachyderm.io avatar

Is ghost on here? Is there a way to beg them not to do this? Random bloggers aren't generally prepared to moderate comments on their own pages. There is no way on earth they're prepared to moderate their subscribers across the whole social web. It's a catastrophically bad idea.

https://activitypub.ghost.org/day2/

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@J12t first, that was kind of a nightmare. That situation gave us the maxim "don't read the comments"

Second, what they're proposing sounds like anyone who signs up for your newsletter can leave comments on any blog, not just yours

jenniferplusplus, to random
@jenniferplusplus@hachyderm.io avatar

mastodon be normal for 5 seconds challenge (difficulty level impossible)

grimalkina, to random
@grimalkina@mastodon.social avatar

Pointed to this paper from a column on it: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=4464593

Folks in dev psych and elsewhere often talk about girls being underconfident. But how rarely we frame in terms of boys' overconfidence.

"Across a range of countries, contexts, and domains, men have been found to exhibit higher degrees of confidence in their ability than women (Kay and Shipman, 2014). This phenomenon has been particularly salient in the fields of science, technology, engineering, and mathematics (STEM)."

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@grimalkina @Di4na That is surprising. Wild guess here, but I wonder to what extent those women's experience with code review is informing how they report code quality.

polotek, to random
@polotek@social.polotek.net avatar

I started down this road because I asked about seeing a full activity log on my own server instance. And it turns out you just can't.

My goal is to make it easy to stand up two separate mastodon instances on my laptop. Then I can get them to talk to reach other and observe what is actually happening. I think that'll teach me a lot

https://social.polotek.net/@polotek/112453817656490883

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@polotek I put together a docker compose sandbox for running fedi services for testing. You might find it useful.

I'm not sure what access you're looking for, but it might be a start. All of the federated traffic in that system should route through a traefik reverse proxy, which might give you a good point to hook in for visibility.

https://github.com/Letterbook/Sandcastles

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@polotek yeah, ssl and host names are just assumed to be present at a lot of levels. Even so, I still had to put mastodon in development mode to get it to be willing to send requests to hosts that resolve to localhost. I assume that's a reaction to a real problem, but I didn't find a simple explanation when I looked.

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@konnorrogers @polotek
It's not a rails thing. Mastodon does it themselves. They went out of their way to do it.

https://github.com/mastodon/mastodon/blob/3837ec2227159d5b05f07d4d9c4a01647f29ad6f/app/lib/request.rb#L330

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@konnorrogers @polotek it's possible that was broken? I saw there's been a recent commit when I was looking it up.

Regardless, it is yet one more thing that makes mastodon hard to build, configure, deploy, and run. Especially for testing.

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@polotek @konnorrogers I assume the major worry is to stop a malicious peer from tricking mastodon into sending activities to its own inbox. But yes, you're right

jenniferplusplus, to random
@jenniferplusplus@hachyderm.io avatar

Do you ever just look at big tech, and think wow, this trajectory looks exactly like big oil, big tobacco, big automotive, and so on?

jenniferplusplus, to random
@jenniferplusplus@hachyderm.io avatar

I really did not want to go back to maintaining a linux desktop as my daily driver, but microsoft just keeps insisting

jenniferplusplus, to random
@jenniferplusplus@hachyderm.io avatar

What the fuck
https://soc.jrconlin.com/@jrconlin/112470747773418801

hazelweakly, to random
@hazelweakly@hachyderm.io avatar

Tell me you're doing the GitOps thing without telling me you're doing the GitOps thing

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@hazelweakly should we tag ourselves? I'm yml gonna yml

hrefna, to fediverse
@hrefna@hachyderm.io avatar

Part of my frustration with #ActivityPub and one of the things I find baffling giving everything else in it: the lack of tools for backpressure.

Backpressure is fundamental in building reliable distributed systems (c.f., Notes on Distributed Systems for Young Bloods). From a C2S perspective I get why it wouldn't need to be specified, but from a S2S federated protocol perspective its absence is frustrating.

All that it says is to take care not to overwhelm others and a bit on rate limits

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@hrefna yes.

But I also want state convergence. No matter what, different hosts are going to wind up with different views of the state of the graph. Why are there no mechanisms to detect that and enable convergence?

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@SoniEx2 @hrefna These things are all fixable problems, it "just" takes coordination between a few dozen part time volunteer implementers 😭

I tell myself I'm going to just do it unilaterally and then help everyone else catch up. I'm not sure if I'm lying to myself or not.

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@julian @mikedev
Fedbox does, and they use the Replies collection for it.

https://metalhead.club/@mariusor/111716353263015161

Letterbook will. I'm flexible as to how, but my current plan is to duplicate the OP's replies collection into the context.

cc @mariusor

gvwilson, to random
@gvwilson@mastodon.social avatar

I want the equivalent of 'lint' or 'ruff' for GitHub repos. Does the repo name match a pattern? Are the right issue labels set up? Et cetera. I can write a script to do this, but if it already exists, a pointer would be welcome: how do you manage the 780 repos your org has?

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

@gvwilson I haven't used it, but there's a github provider for terraform that might help.

jenniferplusplus, to random
@jenniferplusplus@hachyderm.io avatar

gotta say, I'm not at all excited for the next decade of my career to be characterized by fire fighting new prompt injections in places people didn't even think there was a prompt happening

jenniferplusplus,
@jenniferplusplus@hachyderm.io avatar

Like, when 1.5 unpaid volunteer maintainers of log4j have this problem, it's a global industrial disaster and like 10% of all professional dev and/or ops get pulled into at least some portion of the cleanup.

When sam altman and peter thiel spend 100 billion dollars to do the same thing, it's a revolution and the future and anyone who thinks otherwise is an ignorant naysaying doomer.

jenniferplusplus, to random
@jenniferplusplus@hachyderm.io avatar

FYI: you do not, ever, gotta hand it to 'em, if they're fascists. If the worst person you know made a great point, they're still the worst person you know and you need to stop letting them shape the way you think.

  • All
  • Subscribed
  • Moderated
  • Favorites
  • •
  • JUstTest
  • mdbf
  • ngwrru68w68
  • modclub
  • magazineikmin
  • thenastyranch
  • rosin
  • khanakhh
  • InstantRegret
  • Youngstown
  • slotface
  • Durango
  • kavyap
  • DreamBathrooms
  • megavids
  • GTA5RPClips
  • tacticalgear
  • normalnudes
  • tester
  • osvaldo12
  • everett
  • cubers
  • ethstaker
  • anitta
  • provamag3
  • Leos
  • cisconetworking
  • lostlight
  • All magazines
    •