elm-city-craftworks / practicing-ruby-web Goto Github PK

From @sandal:

1. Add a boolean called active_account which is off by default on all users.
2. Set this field to true on login
3. Set up an unsubscribe hook which sets this field to false when people unsubscribe.
4. Only send notification emails to active accounts.

I've begun work on a branch to allow administrators to e-mail the full body of an article to subscribers as and when they are first published (following the example of James Edward Gray II's "Rubies in the Rough").

In order to retain the rich formatting shown on the Practicing Ruby web site, I'm experimenting with a multipart email showing HTML content for those who support it but gracefully degrading to plain text for those without (see CampaignMonitor's "Guide to CSS support in Email re support). It would also make sense to allow users to choose their delivery format in case they wish to receive only the plain text version.

However, to be more representative of actual Practicing Ruby content, could I please have the original Markdown source of a representative article, perhaps one that is already free to the public such as "Unobtrusive Ruby" in Practice" or perhaps one with images such as "Using games to practice domain modeling"?

At the moment, I believe @sandal manually sends announcement emails (please correct me if I'm wrong) so perhaps it would be good to allow for that freeform announcement message to be specified in the mailshot as well (though this could wait).

You know what you doing.

See community for examples.

Clickable and has share link

⭐ ⭐ ⭐ ⭐ ⭐ ⭐
⭐ ⭐ ⭐ ⭐ ⭐ ⭐
⭐ ⭐ ⭐ ⭐ ⭐ ⭐
⭐ ⭐ ⭐ ⭐ ⭐ ⭐
⭐ ⭐ ⭐ ⭐ ⭐ ⭐
⭐ ⭐ ⭐ ⭐ ⭐ ⭐
⭐ ⭐ ⭐ ⭐ ⭐ ⭐
⭐ ⭐ ⭐ ⭐ ⭐ ⭐
⭐ ⭐ ⭐ ⭐ ⭐ ⭐
⭐ ⭐ ⭐ ⭐ ⭐ ⭐
⭐ ⭐ ⭐ ⭐ ⭐ ⭐

Currently if a user is deleted their comments, shared articles, and @ mentions are abandoned. In general users should never be deleted, but it appears there is a case were that has happend (User#id == 53) in production. Now that user's share links are failing.

I see a few ways forward:

1. Delete that user's shared article links and add :dependent => : destroy to the has_many :shared_articles relationship. A similar approach could be taken for comments, and we'd have to do some fancy footwork to invalidate caches where the user was mentioned.
2. Systematically disable the ability to delete users and instead forcing administrators to disable accounts.

Personally I am not too crazy about the first approach. I want those share links to keep working and all user's comments to remain intact. A nice compromise for User #53 might be to assign their share links to @sandal or @jordanbyron.

I reverted the MdMention patch, because the implementation is going to need several changes before it can be used in production:

1. It should check mentions against our users before converting them into links
2. It should work only in comments, not in every markdown field.
3. It should not match things like: [email protected], mentions should only be processed if they are preceeded by the beginning of the line or whitespace.

We had some tests for dealing with mention edge cases (not all of the above were tested, but some were), but this patch completely bypasses them. We'd want proper tests in place for all of the above, as well as anything we originally tested for.

In general, whenever we add something using RedCarpet, it seems like we need to put extra effort into detailed manual testing. Every change made there can potentially break every page on the site, so... more attention to detail is needed. 🚒

Just need to delete one of the links from the page.

This should shut up the horrible bots, or so I hope.

This is mostly a brain dump and will need to be fleshed out before it can be implemented:

• Store articles in markdown files in a separate git repository, which should be symlinked into app.
• Continue to save article meta data (Title, Issue #, Volume #) in the database and load body by matching on Issue # and Volume #
• Add Article#updated_at to the article cache key so it is invalidated when article is updated.
  • We will also need to add template updated at time to this key.
• Syncing article files should update Article#updated_at to the file's date modified time if it is less than Article#updated_at
• Cache Cooker should update any changed articles when they are synced

A user was complaining because they had confirmed their account but then later attempted to log in by clicking the confirmation link. I think it's OK for this to fail, but ideally with a more clear error than "Sorry, something went wrong". Should say something like "That confirmation link has already been used, [try logging in] instead. If you run into problems, contact [email]"

• url
• hashed_ip
• share_key
• user_id
• time_on_page
• start_time
• end_time

Create before_filter. Don't log redirects, webhooks, logins, etc

I read Practicing Ruby on an iPad typically, and as someone who likes to have a clean set of icons, I would love if the bookmark had a robo or ruby icon. To do this, it seems that we just need to create a 512px x 512px png file, similar to: http://everytimezone.com/everytimezone-512.png

The code that they use on that site to make the bookmark have an icon is then:

<link rel=apple-touch-icon-precomposed href=everytimezone-512.png>
<link rel=icon type=image/png href=everytimezone-512.png>

If someone can provide a vector version (or 450ish px version) of robo or the practicing ruby favicon I'd be happy to play around with the image until it looks good as an icon and submit a patch.

Something like the problems page would work well.

Sometimes I accidentally replace old articles with new ones. This is annoying and would be confusing to users. I should display a confirmation whenever editing an article with published status.

Using this as a post-deploy hook might do the trick

https://gist.github.com/122738

This is mostly a brain dump and will need to be fleshed out before it can be implemented:

• Store articles in markdown files in a separate git repository, which should be symlinked into app.
• Continue to save article meta data (Title, Issue #, Volume #) in the database and load body by matching on Issue # and Volume #
• Add Article#updated_at to the article cache key so it is invalidated when article is updated.
  • We will also need to add template updated at time to this key.
• Syncing article files should update Article#updated_at to the file's date modified time if it is less than Article#updated_at
• Cache Cooker should update any changed articles when they are synced

@sandal: I've been playing around with Gollum today and wanted to clarify some assumptions before starting implementation. Here's the requirements I have.

1. Spawn a new Gollum wiki on the VPS.
2. Replicate the articles in the PR database in the wiki
3. Update the PR codebase to use the wiki as a database, pulling articles from it instead of the db.

Please correct or expand on any of these assumptions.

Sometimes it gets put into "unmonitored" mode and doesn't restart.

Just a suggestion - I'm assuming this was done on purpose but maybe you can at least show how many comments the shared article has or let the user know it's possible to comment on the article. It will help users understand practicing ruby better by knowing there's a conversation that takes place and possibly give them more incentive to join.

Backwards compatible with ID based routes

Similar to linking to a selection in a Github source file.

This is a bit of a yak-shave, but easier to do this while there are very few tests.

It's much more complicated than it needs to be, and it looks like it's already got some test coverage, so it's ripe for improvement.

Had to revert the emoji patch because it was breaking a sample in this article:
http://practicingruby.com/articles/1

It was running into a stack trace which contained :2:

If we want to support Emoji on Practicing Ruby, we need to be sure it only runs against text that's not in code block (or any preformatted / blockquote / similar type thing)

I really want to ❤️ this change, but right now it's giving me a 💔

from @brentvatne:

sometimes when i login, already being logged in github, it says the page cannot be found...

He's unable to reproduce but I'm just making this a placeholder because he'll email me next time it happens and I'll check the logs and post back here.