I cannot find where exactly in foreman, but it apparently sets PORT=5000 by default? But why tho? What if I have my own PORT default inside the initialize code? What if my Procfile doesn't start any network services?

also surprised by how many developers don't understand how POSIX setenv(3), unsetenv(3), and fork(2), execv(3) vs execvpe(3) interact. Understanding ye olde POSIX does come in handy from time to time.

I cannot believe rubygems still supports that weird `rails _1.2.3_ ...` argument syntax for which version of an executable gem to load. This is precisely the use-case for a non-exported env variable (you can `ENV.delete` to prevent it from being inherited by child processes).

I have not been able to reproduce this locally, only with these private repos. The mystery deepens...

also when did bundler start pulling in development dependencies of other gems? A development dependency of `rake ~> 10.0` from another gem is conflicting with the development dependency of `rake ~> 11.0` of the main gem I'm updating. I thought development deps were ignored?

Just noticed that `bundle gem` still adds `rake ~> 10.0` as a development dependency, which then . Seems a bit dated?

Update: submitted issues/PRs and both got fixed/merged. See something, say something. Developers can't fix issues they don't yet know about. Even if you decide not to use a library, at least tell the maintainers that something's amiss.

Wasted three days due to outdated [ruby] AWS DynamoDB documentation. Your documentation should have the same degree of coverage and accuracy as your tests.

Eric Pigeon solved the regexp issue that was blocking bundler-audit 0.6.1. Was missing the `!` character...

FYI will be traveling for work, so expect bundler-audit 0.6.1 to be released on/after Thursday.

Proc#<< might get merged into ruby-trunk. Functional style closure composition! Shweet!

One anti-pattern I'm starting to see with GraphQL, is the temptation of getting all of the data in one request. The problem with that mindset is that some of your associated models do not need to be queried/embedded/retransmitted for each request/response. Ideally one would request the shared data, cache it, and join it on the client-side.

So what have people built using Serverless or FaaS Ruby? I'm curious what the limitations/constraints are.

Also I really wish LXD caught on. I like how it gives you the benefits of both containerization and the familiar structure of a GNU/Linux base system (systemd, crond, /bin, /sbin, etc).

Catching up with 2018. Noticing there's a few variations on a stock ruby app Dockerfile. Most use puma, while some use the passenger base image + nginx. Am I the only one who finds it odd how we've gone back to reverse proxying requests to a ruby web server?

Anyone else see the similarities between Rails' Action Text and Shoes, Processing, or Arduino's light-weight editor? Certainly a nice-to-have user-friendly feature, but not a new idea.

github.com/postmodern/digest-c had no clue ruby-kafka was using digest-crc. Looks like I'll be spending some time optimizing digest-crc soon.

Show more

A Mastodon instance for Rubyists & friends