make it work with ruby 3.1 by upgrading thin to the latest version … by Wanderfalke · Pull Request #496 · sj26/mailcatcher

Wanderfalke · 2022-01-27T16:00:49Z

…and replacing skinny with faye-websocket.

also allow mailcatcher process to be stopped with SIGINT / CTRL+C

…and replacing skinny with faye-websocket + allow mailcatcher process to be stopped with SIGINT

Vincent14 · 2022-03-01T13:20:40Z

Hi @sj26 or @tagliala @westonganger could you have a look please? <3

westonganger · 2022-03-02T22:38:45Z

Unfortunately I am not a maintainer of this project. I think I reached out before. But if more maintainership help is needed I am willing.

sj26 · 2022-03-02T23:35:23Z

Thanks for the contribution! But there is already a conflicting stategy for upgrading thin and websockets etc over here:
#467

The blocker is testing, including adding more feature spec coverage. That blocker still exists on this branch. If you're keen to contribute, looking at the surface area of mailcatcher and adding feature specs would be a great contribution.

sj26 · 2022-03-02T23:38:19Z

allow mailcatcher process to be stopped with SIGINT / CTRL+C

This should already be working:

sj26 · 2022-03-02T23:40:18Z

Reading this again, it's a pretty small intervention so might be easier to land. I'll do a review and maybe we can land it.

sj26 · 2022-03-02T23:41:06Z

    # One EventMachine loop...
    EventMachine.run do
+      trap("INT") do
+        EventMachine.add_timer(0) do


I think EventMachine should already be quitting cleanly. But, if not, I think for this line you want:

Suggested change

EventMachine.add_timer(0) do

EventMachine.next_tick do

You're right. The magic was to pass signals: false to thin.

But now thin doesn't exit cleanly?

Sorry, by "EventMachine should already be quitting cleanly" I meant that thin's signal handler is enough to shut down the reactor including draining thin's connections cleanly. Did that break? Is that why you made this change initially?

Ctrl+C now does this for me:

▲ ~/Developer/mailcatcher bundle exec mailcatcher --foreground Wanderfalke-main 9m ● ⬡ Starting MailCatcher v0.8.1 ==> smtp://127.0.0.1:1025 /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/net/protocol.rb:66: warning: already initialized constant Net::ProtocRetryError /opt/rbenv/versions/2.7.5/lib/ruby/gems/2.7.0/gems/net-protocol-0.1.2/lib/net/protocol.rb:68: warning: previous definition of ProtocRetryError was here /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/net/protocol.rb:206: warning: already initialized constant Net::BufferedIO::BUFSIZE /opt/rbenv/versions/2.7.5/lib/ruby/gems/2.7.0/gems/net-protocol-0.1.2/lib/net/protocol.rb:208: warning: previous definition of BUFSIZE was here /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/net/protocol.rb:503: warning: already initialized constant Net::NetPrivate::Socket /opt/rbenv/versions/2.7.5/lib/ruby/gems/2.7.0/gems/net-protocol-0.1.2/lib/net/protocol.rb:504: warning: previous definition of Socket was here ==> http://127.0.0.1:1080 ^CTraceback (most recent call last): 20: from /opt/rbenv/versions/2.7.5/bin/bundle:23:in `<main>' 19: from /opt/rbenv/versions/2.7.5/bin/bundle:23:in `load' 18: from /opt/rbenv/versions/2.7.5/lib/ruby/gems/2.7.0/gems/bundler-2.1.4/libexec/bundle:34:in `<top (required)>' 17: from /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/bundler/friendly_errors.rb:123:in `with_friendly_errors' 16: from /opt/rbenv/versions/2.7.5/lib/ruby/gems/2.7.0/gems/bundler-2.1.4/libexec/bundle:46:in `block in <top (required)>' 15: from /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/bundler/cli.rb:24:in `start' 14: from /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/bundler/vendor/thor/lib/thor/base.rb:476:in `start' 13: from /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/bundler/cli.rb:30:in `dispatch' 12: from /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/bundler/vendor/thor/lib/thor.rb:399:in `dispatch' 11: from /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/bundler/vendor/thor/lib/thor/invocation.rb:127:in `invoke_command' 10: from /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/bundler/vendor/thor/lib/thor/command.rb:27:in `run' 9: from /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/bundler/cli.rb:476:in `exec' 8: from /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/bundler/cli/exec.rb:28:in `run' 7: from /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/bundler/cli/exec.rb:63:in `kernel_load' 6: from /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/bundler/cli/exec.rb:63:in `load' 5: from /opt/rbenv/versions/2.7.5/lib/ruby/gems/2.7.0/bin/mailcatcher:23:in `<top (required)>' 4: from /opt/rbenv/versions/2.7.5/lib/ruby/gems/2.7.0/bin/mailcatcher:23:in `load' 3: from /Users/sj26/Developer/mailcatcher/bin/mailcatcher:6:in `<top (required)>' 2: from /Users/sj26/Developer/mailcatcher/lib/mail_catcher.rb:180:in `run!' 1: from /opt/rbenv/versions/2.7.5/lib/ruby/gems/2.7.0/gems/eventmachine-1.0.9.1/lib/eventmachine.rb:193:in `run' /opt/rbenv/versions/2.7.5/lib/ruby/gems/2.7.0/gems/eventmachine-1.0.9.1/lib/eventmachine.rb:193:in `run_machine': Interrupt

My initial motivation for this was the fact, that for me on MacOS 12.1 (Intel) CTRL+C / SIGINT did nothing and mailcatcher process can only be stopped with SIGKILL.
Passing "signals: false" to thin makes SIGINT work, but as you state above also prints stacktrace. Explicitely trapping INT signal in the eventmachine loop makes it exit cleanly.
BTW, your suggestion to use "EventMachine.next_tick" leads to following error: eventmachine.rb:1117:in "synchronize": can't be called from trap context (ThreadError)

sj26 · 2022-03-02T23:42:47Z

-  s.add_dependency "thin", "~> 1.5.0"
-  s.add_dependency "skinny", "~> 0.2.3"
+  s.add_dependency "thin", "~> 1.8.0"
+  s.add_dependency "net-smtp", "~> 0.3.0"


Why was net-smtp added here?

net-smtp

net-smtp has been extracted from the default gems in Ruby 3.1

Ref:

Temporarily add net-gems as dependencies of frameworks that use mail rails/rails#44083

https://www.ruby-lang.org/en/news/2021/12/25/ruby-3-1-0-released/

As already mentioned by @tagliala, as of Ruby 3.1, net-smtp was moved to bundled gems.

Ah, I see. Thanks.

This now spits a lot of warnings during startup for me on Ruby 2.7.5:

$ bundle exec mailcatcher --foreground Starting MailCatcher v0.8.1 ==> smtp://127.0.0.1:1025 /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/net/protocol.rb:66: warning: already initialized constant Net::ProtocRetryError /opt/rbenv/versions/2.7.5/lib/ruby/gems/2.7.0/gems/net-protocol-0.1.2/lib/net/protocol.rb:68: warning: previous definition of ProtocRetryError was here /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/net/protocol.rb:206: warning: already initialized constant Net::BufferedIO::BUFSIZE /opt/rbenv/versions/2.7.5/lib/ruby/gems/2.7.0/gems/net-protocol-0.1.2/lib/net/protocol.rb:208: warning: previous definition of BUFSIZE was here /opt/rbenv/versions/2.7.5/lib/ruby/2.7.0/net/protocol.rb:503: warning: already initialized constant Net::NetPrivate::Socket /opt/rbenv/versions/2.7.5/lib/ruby/gems/2.7.0/gems/net-protocol-0.1.2/lib/net/protocol.rb:504: warning: previous definition of Socket was here

If it's loading stdlib and the gem, that might cause issues? How does one avoid this double loading on older rubies?

Tried ruby 2.7.5 and I'm getting same warnings. Not sure how to avoid that yet, will keep looking for a solution.

Following this issue: ruby/net-imap#16 , adding net-http as dependency solves "already initialized constant" issues with net-protocol.

Wanderfalke · 2022-03-04T16:14:33Z

@sj26 Thank you for taking time to review my PR. I addressed your comments. Please have another look when you can.

sj26 · 2022-03-05T01:01:22Z

I've added a spec on main which tests that mailcatcher exits cleanly on interrupt, and rebasing this branch to include that spec causes a failure:

  1) Quit quits cleanly on Ctrl+C
     Failure/Error: expect(status).to be_exited
       expected `#<Process::Status: pid 56232 SIGINT (signal 2)>.exited?` to be truthy, got false
     # ./spec/quit_spec.rb:41:in `block (2 levels) in <top (required)>'

shaneshort · 2022-05-27T02:48:18Z

any particular reason this wasn't merged? I'm currently unable to install this on ruby 3.x and I suspect this PR would help.

sj26 · 2022-05-27T02:55:56Z

There is a failing spec rebased on main, and not enough testing generally to be sure it hasn't broken anything.

sj26 · 2022-07-04T00:51:30Z

Hi folks! Thanks for this work here, especially @Wanderfalke.

Ruby 3.1 support and a switch to faye-websocket are available in 0.9.0.beta1. I've love some folks to test. You can install the latest pre-release with:

gem install --pre mailcatcher

I think there are still some quirks. I don't think the websockets send the quit message before shut down. But that's not critical.

I'm keen to upgrade Thin, too, but I'd like to bed in the switch to faye, first. Beyond that I'll see if we can complete the switch to the async gems (#467).

larryh · 2023-08-04T18:52:32Z

I know this is marked as Closed (so hopefully I'm not wasting my time typing this), but I wanted to give the maintainers and anyone else who has this problem some feedback...

I'm using Rails 3.1.0, and when I installed and tried to run mailcatcher 0.9.0 this is what I got:

Unable to load the EventMachine C extension; To use the pure-ruby reactor, require 'em/pure_ruby'
:85:in `require': libssl.so.1.1: cannot open shared object file: No such file or directory - /home/larry/.rbenv/versions/3.1.0/lib/ruby/gems/3.1.0/gems/eventmachine-1.0.9.1/lib/rubyeventmachine.so (LoadError)
        from :85:in `require'
        from /home/larry/.rbenv/versions/3.1.0/lib/ruby/gems/3.1.0/gems/eventmachine-1.0.9.1/lib/eventmachine.rb:8:in `'
        from :85:in `require'
        from :85:in `require'
        from /home/larry/.rbenv/versions/3.1.0/lib/ruby/gems/3.1.0/gems/mailcatcher-0.8.1/lib/mail_catcher.rb:7:in `'
        from :85:in `require'

So I ran the following two commands:

gem uninstall mailcatcher
gem install --pre mailcatcher

And it worked like a charm, specifically:

larry@system76-pc:~/RubymineProjects/BackEnd$ mailcatcher
Starting MailCatcher v0.9.1.beta1
==> smtp://127.0.0.1:1025
==> http://127.0.0.1:1080
*** MailCatcher runs as a daemon by default. Go to the web interface to quit.

Thanks for the great gem!

sj26 · 2023-08-25T23:33:42Z

Sorry @larryh, that's very weird!

That sounds like a problem within eventmachine. Perhaps it linked to an older version of openssl? Are you using homebrew? Or another system package manager? It might have upgraded shared libraries beneath your feet.

Reinstalling the gems usually fixes it up for me:

gem pristine eventmachine
# or
gem pristine --all

larryh · 2023-08-26T14:05:29Z

Hi,
Thanks for the constant feedback - I know you must be busy and it is much appreciated.

A "gem list" command tells me I have Eventmachine 1.2.7, which I believe is the latest version.(From 5 years ago!) It's not in my (Rails project) Gemfile or even Gemfile.lock. Also, I don't think I ever intentionally installed it, so maybe it just comes along for the ride when I install Mailcatcher.

Also, I'm not on a Mac, so no homebrew. (I'm on Ubuntu)

I'll just continue to refresh the page to see new emails.

Thanks again,
Larry

sj26 · 2023-08-27T00:45:39Z

Heh, Homebrew runs on Linux these days.

But even apt may have bumped your libssl. And apt isn't aware of rubygems, so wouldn't cause dependent gems to be recompiled.

I'd recommend a gem pristine --all just to be safe :-)

larryh · 2023-08-28T12:46:36Z

Thanks for the tip. I did a "gem pristine eventmachine" but that didn't do anything; I still have to refresh the page to see the email.

I don't want to try the "all" command because I'm gun-shy of potential unintended consequences. Oftentimes (or at least sometimes) issuing commands that sweeping can result in spending hours just trying to get back to where you were, i.e. "when things worked."

I'll just stick with refreshing the page when I need to.

Thanks for all of your help!

+ make it work with ruby 3.1 by upgrading thin to the latest version …

eb03fdf

…and replacing skinny with faye-websocket + allow mailcatcher process to be stopped with SIGINT

Wanderfalke mentioned this pull request Jan 27, 2022

Mailcatcher install fails on ruby 3.1.0 #495

Closed

sj26 reviewed Mar 2, 2022

View reviewed changes

Comment thread lib/mail_catcher/web/application.rb Outdated

sj26 reviewed Mar 2, 2022

View reviewed changes

Comment thread lib/mail_catcher/web/application.rb Outdated

sj26 reviewed Mar 2, 2022

View reviewed changes

code review changes

f1e379d

Include testing ruby 3.1 on CI

c17f876

Wanderfalke and others added 9 commits March 6, 2022 14:34

Merge branch 'sj26:main' into main

74f97dd

bring back INT signal handling

f71d348

add net-http as dependency constant init warnings

9140a60

trigger ci

95872aa

Create main.yml

4c07d57

remove workflow

b8197d1

try latest bundler

a122c04

fix workflow syntax

13294a0

fix workflow syntax

544578c

Wanderfalke closed this May 8, 2022

sj26 mentioned this pull request Jul 3, 2022

Switch from Skinny to Faye Websocket #510

Merged

sj26 mentioned this pull request Jan 25, 2023

Update thin version #530

Merged

larryh mentioned this pull request Aug 8, 2023

No longer updates just-sent emails #550

Open

Conversation

Wanderfalke commented Jan 27, 2022

Uh oh!

Vincent14 commented Mar 1, 2022

Uh oh!

westonganger commented Mar 2, 2022

Uh oh!

sj26 commented Mar 2, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

sj26 commented Mar 2, 2022

Uh oh!

sj26 commented Mar 2, 2022

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

sj26 Mar 4, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Wanderfalke commented Mar 4, 2022

Uh oh!

sj26 commented Mar 5, 2022

Uh oh!

shaneshort commented May 27, 2022

Uh oh!

sj26 commented May 27, 2022

Uh oh!

sj26 commented Jul 4, 2022

Uh oh!

larryh commented Aug 4, 2023

Uh oh!

sj26 commented Aug 25, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

larryh commented Aug 26, 2023

Uh oh!

sj26 commented Aug 27, 2023

Uh oh!

larryh commented Aug 28, 2023

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

sj26 commented Mar 2, 2022 •

edited

Loading

sj26 Mar 4, 2022 •

edited

Loading

sj26 commented Aug 25, 2023 •

edited

Loading