Notices by Erin 💽✨ (erincandescent@akko.erincandescent.net), page 6

I love sending email to GMail users :akko_sad:

The more I think about it, the more I feel that Postel’s law (“be conservative in what you send, be liberal in what you accept”) is an aphorism that is very applicable to file formats and much less applicable to protocols

With protocols you have the opportunity to reject then and there; with file formats, you generally just have to deal with whatever garbage someone emitted 30 years ago and in general you can’t know who the recipient is so all content adaptation has to be done on the recipient side (or by the sender providing graceful degredation alternatives)

Anyway I don’t know how much they’re actually paying (or would actually be paying) for bandwidth, but in context of them paying >$1000/mo for compute it’s not beyond reason

$3600/mo or so if you store it all in S3, of course

I saw someone saying Hachyderm’s egress traffic is 1TB a day. So, dear internet, I wonder: how much would you expect serving 1TB/day (~30TB/month) to cost, just in bandwidth costs?

Delivered on a fiber cable to you: almost certainly less than €150/mo. You could get down lower with a sufficiently friendly carrier.

Some VPS providers will do much cheaper but you may be gambling a little on network quality pushing that much

@ireneista @aeva @CliftonR For better and for worse, its our understanding that if you properly implement and handle the List-Unsubscribe header most of the bigass e-mail providers won’t classify your email as spam too.

For better because most mail clients now have an “Unsubscribe” button on marketing emails and it actually works. For worse because they don’t get autofiled into your spam box for you

@ireneista @CliftonR @aeva (Also so much of the marketing crap is coming from Mailchimp/Sparkpost/a handful of others these days so you can get rid of so much of it by filtering list IDs ending .mcsv.net or sparkpostmail.com)

@CliftonR @aeva @ireneista Sometimes it feels like I receive more marketing e-mails disguised as service-related transactional e-mails these days than I received actual spam 10 years ago.

Sigh.

@CliftonR @aeva @ireneista I do not blame anyone for having a disaster inbox; keeping things under control is hard. I recently learned Fastmail lets me do address+folder.path@domain.com, which helps shovel the transactional e-mails off to the side pre-emptively (everything gets a unique e-mail address starting with +bulk. so that if there isn’t a dedicated folder for it, it gets shoved into the Bulk folder)

But it really shouldn’t be like this. It shouldn’t be this hard to keep control of one of your primary communications channels.

@lanodan I have lore.kernel.org added to Thunderbird via NNTP and its great.

Its deeply upsetting that the specification for mailto:address?in-reply-to=message_id is 26 years old and yet barely any software supports it besides Thunderbird :-(

And the alternative option is downloading an e-mail message file (.eml), opening it, and replying to that; but how many people even have desktop e-mail clients setup any more?

I understand the dislike for e-mail based patch workflows. I have a soft spot for sending my changes with git send-email but I don’t think its a good code review or change management process.

But mailing lists… I don’t really understand the dislike people have for mailing lists. Some of the best technical collaboration I’ve ever done has been via mailing lists. Most of them you don’t even have to be a subscriber to start conversations on; they’ll accept e-mails from anyone. And, unlike with forums, the replies come back to my e-mail inbox, which means its harder to forget about posts I’ve made.

Of course there is something to be said about how e-mail tooling can be obnoxiously minimalist, arcane, or just a thing that people don’t have setup any more in the same way as they did in the past.

@pid_eins I was about to say “we’re getting PIDFD_SELF and you could use that in the same way as e.g. AT_FDCWD“ except both PIDFD_SELF and AT_FDCWD are defined as -100. This sucks. Maybe we could get them made into different numbers before 6.13 drops? ;_;

@pid_eins My perhaps controversial opinion is that from userland the magic file descriptors should Just Work like actual file descriptors in every regard except that you can’t close() them or dup2 to them, but I guess that ship has sailed anyway.

(i.e. you should be able to stuff PIDFD_SELF into an SCM_RIGHTS control message and out of the other end pops a pidfd for your process; no need to pidfd_open(getpid()) and close() it)