According to Discord, Go was (or at least it used to) cause huge latency spikes because of it's GC - they've rewritten it in Rust and now it's A LOT faster, without any latency spikes.
When does one decide whether an environment is a test? I would like to test SMTP notifications of a staging server at work, approximately 15 million emails to be sent. This is NOT production, but those 300 ms spikes would be unacceptable. That Rust rewrite seems to be a godsend.
The GC iterations they were talking about were during incredibly high throughout. Most users would not notice this. It's not some general indictment of Go.
I don't know if the author is on here, but I wonder what it was like doing the frontend in Rust (Yew). How does the developer experience compare to Vue, I suppose?
Hi, the author of MailCrab here :-) Yew is nice, especially if you enjoy writing Rust. However, it definitely takes more time and dedication than writing a frontend in React, Vue etc. Yew and the surrounding ecosystem keeps improving, and it is way more usable than when I first tried it. The tooling I used (Trunk https://trunkrs.dev/) is very minimal with respect to the number of features compared to many of the popular web-bundlers (Webpack etc.) but it works well for most simple use-cases.
The terminal solution is handy but sometimes it fails because you could end up with a sign in link or some type of link with a very long token that wraps to multiple lines making it impossible to click.
Rails also prints emails to the terminal and has a backend solution that doesn't really send emails out but I still prefer using MailCatcher to display them in a UI (it has both amd64 and arm64 Docker support[0], real-time updates and a ~25mb compressed Docker image).
It also just works for all frameworks if you hack the receiver and BCC: emails you want to test to yourself to make sure they're actually being sent and look alright.
That’s nice since MailHog seems dead. I’ve had PRs open for their SMTP server package for years. One for a bug, the other to support Unicode recipient addresses.
Once I got one of their attention on Twitter and they said they would have it looked at the next week, but nothing seemed to happen.
I have used the Java smtp test email server FakeSMTP which has been around since around 2015. It can run via CLI as a service with no GUI, or with a GUI interface via Swing:
Longtime mailhog user here; this looks like a nice improvement for cold start developer environments! Another option in the same space (emphasis container size) is https://github.com/inbucket/inbucket
It's finding systems like this that make HN valuable from time to time. I didn't even "think" to look for a system like this, and I could use it right now, and I find there are multiple other ones similar in the same posting from comments.
Looking forward to trying this out. I do a lot of email testing locally for paced.email and vend.email. MailHog is 90% of the way there, but seems like it's fallen by the wayside...
As someone who recently moved to NL and is trying to learn Dutch, I've found that translating plurals is probably the most annoying part of my life at the moment.
"De golven" is the plural form, so "golf" makes sense, but Google Translate doesn't really know that. It did suggest "golven" in the "See also" section, but it wasn't obvious at first. Especially when the sport golf is also spelled the same way!
Other things I have trouble with is sentence structure, but that's not really annoying so much as it is frustration bourn of ignorance.
Which sounded like a covid thing to me. Must have been interesting to watch the access logs for tweedegold.nl in fall 2020! (That is, I assume the name was chosen before covid but the About page doesn't say.)
[0]: https://github.com/axllent/mailpit