Some cleanup #145

pirj · 2020-05-22T22:31:08Z

Ditch puma and foreman

Having Foreman for a single process is pointless. Puma is webscale, but
middleman server works quite fine without it.
Rearrange dependencies

[x] moved bourbon, neat, and slim under dev, only needed to build
[x] removed hard version specifiers - they are locked anyway
[x] bumped builder to get rid of ::Fixnum deprecation warning
[x] changed middleman-syntax to use the released rubygems version
[x] added rack-contrib, no idea how try_static worked without it
[x] bumped middleman-livereload to get rid of some warning
[-] bump middleman* to work with Bundler 2 (middleman 3.x can't) - middleman-deploy is an "old" unsupported extension, and there's no clear path to update. Also source_dir is not defined anymore
Ditch docs-template gem - unused?
Ditch bower

pirj · 2020-05-23T22:22:10Z

source/layouts/layout.slim

@@ -66,5 +66,4 @@ html.no-js lang="en"
        br
        = image_tag "https://cdn.dnsimple.com/assets/resolving-with-us/logo-dark.png", alt: "DNSimple", style: "width:100px;"
    = partial 'google_analytics'


It was also tempting to remove this. Not sure if there's any use for analytics 🔍

JonRowe · 2020-05-24T09:16:36Z

source/_seed_and_dew.html.erb

-    (document.getElementsByTagName('head')[0] ||
-    document.getElementsByTagName('body')[0]).appendChild(snd);
-  })();
-</script>


If we wish to stop our participation in seed and dew thats a discussion we can have, but whilst we're still setup with them and open collective receives money from them on our behalf this needs to remain to track that.

Oh, my bad. I got confused by this line in the console:

Loading failed for the <script> with source “https://www.seedanddew.com/static/embed.min.js”.

Might be my adblocker.
Where do those support buttons are supposed to appear? Can't see anything even in Incognito.

JonRowe · 2020-05-24T09:27:59Z

source/javascripts/docs/app.js

-  snd.src = 'https://www.seedanddew.com/static/embed.min.js';
-  (document.getElementsByTagName('head')[0] ||
-  document.getElementsByTagName('body')[0]).appendChild(snd);
-})();


JonRowe · 2020-05-24T09:54:07Z

bower.json

-  "version": "0.0.0",
-  "homepage": "https://github.com/rspec/rspec-website",
-  "authors": [
-    "Andrew <[email protected]>"


Andrew should be added to the credits somewhere really!

JonRowe · 2020-05-24T09:54:22Z

bower.json

-    "tests"
-  ],
-  "dependencies": {
-    "normalize-css": "~2.1.3"


I think this is still used no?

We do, inline:

source/stylesheets/_application.css.scss 1://= normalize 3:@import "normalize"; source/stylesheets/_normalize.css 1:/*! normalize.css v2.0.1 | MIT License | git.io/normalize */

It seems we're far behind the current v8.0.1 :D

JonRowe · 2020-05-24T09:54:38Z

Gemfile.lock

@@ -170,17 +161,16 @@ PLATFORMS
 DEPENDENCIES
  bourbon
  builder (~> 3.0)
-  foreman
+  bundler (~> 1.17.3)


There is a version of middleman that supports bundler 2

I wasn't prepared for cleanup & full update, but if you ask for it, sure :D

After the update to Middleman 4:

/Users/pirj/.rvm/gems/[email protected]/gems/middleman-core-4.3.6/lib/middleman-core/extensions.rb:96:in `load': Tried to activate old-style extension: deploy. They are no longer supported. (RuntimeError)

I don't see a clear way to deploy otherwise, might be git push staging master would work.
Also, source_dir in RSpecInfo::Helpers.rspec_documentation_latest(source_dir) is not defined, so probably version fallback is broken.

I'd say there's a lot of work with no tangible benefit. Let's stick with Bundler 1 for now?

@JonRowe Can you please grant me permissions to push to https://github.com/rspec-staging/rspec-staging.github.io/ 's master?

I intend to rework the deployment in the following way:

update middleman and middleman-core to the latest

drop middleman-deploy and its configs

use Bundler 2

make build directory a submodule of this very repository (and will make a note to avoid git initialize --recursive just in case, even though the submodule will be on master with no submodules defined)

add info how to add two remotes to that submodule, itself and rspec-staging

to publish: prod (cd build; git push prod master), staging (cd build; git push staging master)

forget about the staging trick of omitting CNAME, if a "Custom Domain" is not filled in in rspec-staging/rspec-staging.github.io/, it should not redirect to the domain then

I'd love to test how this will work on staging

Done, but bare in mind that I'm not sure submodules work with github pages...

It looks like middleman 4.0 update is a major PITA.
The documentation works quite ok
I'm stuck with an endless list of SASS failures:

== Request: /stylesheets/_application.css error build/stylesheets/_application.css Error: File to import not found or unreadable: bourbon. on line 5:1 of source/stylesheets/_application.css.scss >> @import "bourbon"; ^ ... == Request: /stylesheets/_application.css identical build/stylesheets/_code.css error build/stylesheets/_application.css Error: File to import not found or unreadable: bourbon. on line 5:1 of source/stylesheets/_application.css.scss >> @import "bourbon"; ^ /Users/pirj/source/rspec-dev/repos/rspec.github.io/source/stylesheets/_application.css.scss:5

Besides that, I keep finding new things I'd love to improve, namely ditch neat which is not maintained anymore, and it's been recommended to use flexbox instead.

Even with the current dependencies we're getting quite a lot of:

WARNING: The button mixin is deprecated and will be removed in the next major version release

I don't really feel like fixing all of that at once.

Good to merge as is and address middleman 4/neat removal/css fixes separately?

Middleman depend on Bundler 1.x and `bundle` fails if Bundler 2.x is installed, even if Bundler 1.x is also installed.

- moved puma and foreman to dev, only needed to locally run - moved bourbon, neat, and slim under dev, only needed to build - removed hard version specifiers - they are locked anyway - bumped builder to get rid of ::Fixnum deprecation warning - bumped foreman - changed middleman-syntax to use the released rubygems version - added rack-contrib, no idea how try_static worked without it - bumped middleman-livereload to get rid of some warning

When RACK_ENV was not present: missing argument: -e

Having Foreman for a single process is pointless. Puma is webscale, but `middleman server` works quite fine without it.

JonRowe

I tested this still renders with a clean build so it looks ok. I have reverted the bundler changes, bundler 2 works fine with this repo, it just ignores the bundler 1.1 dependency and carries on.

pirj · 2020-07-30T16:21:21Z

Awesome, thanks!

@benoittgt, good to merge?

benoittgt · 2020-07-30T22:14:30Z

Yep ! Thank you very much

…

On July 30, 2020 6:21:39 PM GMT+02:00, Phil Pirozhkov ***@***.***> wrote: Awesome, thanks! @benoittgt, good to merge? -- You are receiving this because you were mentioned. Reply to this email directly or view it on GitHub: #145 (comment)

-- Sent from my Android device with K-9 Mail. Please excuse my brevity.

pirj self-assigned this May 22, 2020

pirj requested review from JonRowe and benoittgt May 22, 2020 22:31

pirj commented May 23, 2020

View reviewed changes

JonRowe reviewed May 24, 2020

View reviewed changes

pirj force-pushed the cleanup branch 2 times, most recently from b8d1256 to 578edc3 Compare May 24, 2020 15:59

pirj added 10 commits July 27, 2020 00:23

Ditch bower

9e86491

Ditch docs-template gem

b04d334

Skip building on startup

b2a81dd

Cosmetically improve docs

6072655

Downgrade bundler

4a561da

Middleman depend on Bundler 1.x and `bundle` fails if Bundler 2.x is installed, even if Bundler 1.x is also installed.

Fix Procfile for missing env vars

471401a

When RACK_ENV was not present: missing argument: -e

Ditch puma and foreman

32fccc3

Having Foreman for a single process is pointless. Puma is webscale, but `middleman server` works quite fine without it.

Add Andrew Harvey to Credits section

12ed896

Bring Bundler 1.17.3 back to Gemfile.lock

1945a7c

pirj force-pushed the cleanup branch from 578edc3 to 1945a7c Compare July 27, 2020 23:09

Manually override bundler version

63e27e4

JonRowe approved these changes Jul 30, 2020

View reviewed changes

pirj merged commit 90a4325 into source Jul 30, 2020

pirj deleted the cleanup branch July 30, 2020 22:40

pirj mentioned this pull request Jan 31, 2021

Upgrade middleman to version 4 #163

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Some cleanup #145

Some cleanup #145

pirj commented May 22, 2020 •

edited

Loading

pirj May 23, 2020

JonRowe May 24, 2020 •

edited

Loading

pirj May 24, 2020

JonRowe May 24, 2020

JonRowe May 24, 2020

JonRowe May 24, 2020

pirj May 24, 2020

JonRowe May 24, 2020

pirj May 24, 2020

pirj May 24, 2020

pirj Jul 26, 2020

JonRowe Jul 27, 2020

pirj Jul 27, 2020

JonRowe left a comment

pirj commented Jul 30, 2020

benoittgt commented Jul 30, 2020 via email

Some cleanup #145

Some cleanup #145

Conversation

pirj commented May 22, 2020 • edited Loading

Choose a reason for hiding this comment

JonRowe May 24, 2020 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

JonRowe left a comment

Choose a reason for hiding this comment

pirj commented Jul 30, 2020

benoittgt commented Jul 30, 2020 via email

pirj commented May 22, 2020 •

edited

Loading

JonRowe May 24, 2020 •

edited

Loading