Katie Hunter wrote:

Hi, I am not sure what these right topic means but i realize that some threads have a very large list and style wise it looks awkward. I even see the last title collapsing over another txt

I think it will look better if we have something like this or better if you click on "Read more" instead of the list expending, because it will make the topic look weird style wise, we can have some sort of a pop up window that has the list of topics/posts.

Posts: 6

Participants: 3

Read full topic

Kane York wrote:

I cannot, for the life of me, figure out where the /read and /posted routes are defined in the Ember app. I'm trying to find out so that I can document what they do for the usage guide, and I'm at an utter loss as to what /posted is supposed to do. As far as I can tell, it's identical to /latest.

Pasted image1154x368 91.6 KB

Posts: 1

Participants: 1

Read full topic

awesome robot wrote:

I hate to constantly reference NodeBB when talking about Discourse improvements because I don't think their product is particularly better overall — but a lot of the feedback I've seen can be boiled down to the idea that NodeBB is really fast.

The comparison leads me to wonder if NodeBB is really only appearing to be faster (I'm not even sure if it's actually faster).

There's definitely room for subjectivity, but I think some thoughts around the perception of speed from Steve Sounders can be relevant here.

The way topics and topic lists are loaded for example... Discourse isn't loading an entire new page like a traditional website may, but it does often feel like it is. You click a topic link, and you immediately get a flash of a white page and a loading indicator, which then flashes over to the content once it's loaded. Logically this is the fastest way to do things.

NodeBB on the other hand pads out their loading through a few fade effects (content fades out and new stuff fades in). I think they're over-animating it... but you're definitely looking at a blank screen or loading indicator less (or not at all), and it feels like you're not waiting as long... you're not being sent to a waiting room to watch a clock.

From a purely front-end perspective I think reducing the visual importance of the current loading indicator (the black box with the spinner) can help ease minds a little, but I was also curious to see if anyone else had related thoughts...

Could it feel better to overlay a loading indicator on the current page rather than a blank white page?

Posts: 7

Participants: 5

Read full topic

basj wrote:

Hi Discourse developers and users,

I recently discovered Discourse, and am really impressed about how fresh / goodlooking it is. I really would like to install it.
(Note : I currently use solutions like Wordpress + PhpBB for my websites.)

An important thing about what made PhpBB one of the biggest forum solution of the last decade was :

• easy install : we only needed PHP, MySQL, and it was it quickly installed !
• it can be installed even on "shared hosting" : lots of people (like me) only have shared hosting and no personal server : this means that we cannot install things like Rails, and cannot install Postgres, etc. (I don't have root access on my shared hosting)

For these reasons, it is now impossible to install Discourse on "shared hosting". I don't have figures showing how many % of website-makers have a dedicated server and how many % have a shared hosting, but if you want Discourse to become a widely used solution, you have to think about this. (I do know you already are thinking about this by the way...)

What are you ideas in order to make Discourse available for more people (like us, shared-hosting users)?

Best regards.

Note : I already have read related topics like https://meta.discourse.org/t/how-to-install-discourse-on-a-shared-host-server/6274/2, https://meta.discourse.org/t/isn-t-there-more-easier-way-to-install-discourse/6794

Posts: 9

Participants: 6

Read full topic

Sam Bauch wrote:

Continuing the discussion from Shouldn't you be able to tag with multiple categories?:

I agree with @weirdcanada that the categorization system works well for thematic organization but that we need a system of tagging. A trite example that illustrates how I'm thinking about it is categories of Fruit and Vegatable, with a goodreads tag. "Good Reads" wouldn't be a great category because it isn't really thematic, and then confuses users who want to share a great article about vegetables (where would it go?)

An idea I had was to implement Twitter-style hashtags, much in the same way Discourse @mentions function.

So a user could tag their post anywhere in the post body with a hashtag (which brings up the autocomplete with existing/popular tags).

It's then rendered as a "chip" in their post, and other users can click that chip to be taken to a search results view that shows all of the posts with that tag. It would also be searchable via the existing search functionality for free.

What do others think of this idea? I think it would be much easier to implement via a PR than it would as a plugin, so I thought I'd get some ideas flowing before starting any work on it.

Posts: 53

Participants: 24

Read full topic

Kane York wrote:

Currently, clicking on an @mention is an 'ejector seat' - you're taken out of the topic to the user's activity page. We already have the 'safety net' with usercards - why not put them on the @mentions too?

This was brought up before: https://meta.discourse.org/t/show-mini-user-profile-popup-when-clicking-on-mentions-or-avatars-within-an-expanded-reply-thread/10631

I'm not quite sure how to write the code to do this, though. I can't add an ember action in mention_dialect.js (I think?) because it isn't Handlebars and I don't have a reference to the user's Ember object.

Also, what file should I write it in? I have no idea.

Posts: 6

Participants: 3

Read full topic

Sebastian wrote:

Considering using DC for a forum in Thailand.

Im not a coder and these questions might seem stupid:

1. Is there a language file I can just send to gengo.com and re upload translated to server where all text is translated, including UI text like buttons etc?

2. Any rough idea about how many words would need to be translated?

3. And the biggest issue. What about when DC adds a new button or text string somewhere. How will I know when to update the Language file? Do you have to monitor for pulls in github or whatever its called?

4. Is the UI adapted for languages that read from right to left as well?

Posts: 3

Participants: 3

Read full topic

FichteFoll wrote:

I did mention this about a year ago, but it was inside of another thread and not the core topic, even though it deserves to be one.

Using this topic as an example, you can see a bunch of "moderative posts" (which is what I'll call them) that tell you that something has been done. They look like normal posts and they are written in the first person perspective, but they are not normal posts, and I feel like the yellow background is not enough to tell that "action has taken place". ^{It was worse when the background was a slight blue.}

So, why is this a problem? Well, I currently see two issues with this:

1. As I saw these the first time I immediately thought: "Why do these people create such a lengthy post for closing a thread?" I know it's common practice to announce who closed a thread so that users know where to address their complaints (I moderated a community for a while), but in that case this is a lot redundant information.
2. It looks like a normal post and not like something created automatically. You can even like it or bookmark it - or flag and reply to it. I guess being able to reply is reasonable since it opens a public discussion about a moderator decision, but all the other stuff is unnecessary and even that is not really useful in 95% of the cases.

As an example for where this is handled differently I'd like to point to github's issues. (Random example issue)

As you can see, action is clearly different from comments and it is immediately obvious what action has been performed due to the icons.
In a place dedicated to discussions (such as discourse) comments and action should be visually distinguishable even for first-time visitors.

Posts: 16

Participants: 8

Read full topic

FichteFoll wrote:

By this moderator action we can see that 11 posts have been moved to create a new topic. When going to that topic, and more specifically to this post, which quotes a post that has been moved along with it, we can observe the following:

The quoted reply is clearly the post above. However, when clicking the arrow to go to the origin of the quote we are redirected to the original topic where the posts have been moved from. However, the post is not there. Furthermore, the link list on the right side also specifies the old topic which can only come from the quote because there are not other links in the post.

Posts: 1

Participants: 1

Read full topic

FichteFoll wrote:

This applies to anywhere inside the thread, but I chose this example because it's very clear to see the issue.

So, the example topic I will use here is this one. By opening the site and having a certain window size that exceeds some threshhold (I changed the window size before reloading), I get to see this:

2014-03-22_05.38.18.png1126x790 98.5 KB

Notice how I can only see one post but the "current post tracker" (or whatever that is called, it didn't reveal its name upon hovering) shows I have already read 3 posts of 89. This continues to be offset by roughly 500-700px for the whole page and happens in every topic I have encountered so far.

The offset size made me think that the whole box after the initial post is not considered in the calculation, but collapsing it did not solve the issue. I am still told to be reading the 6th post despite only being able to see the first 4.

As you probably know, I'm on FireFox (currently 29 beta but this bug has existed for a longer time).

Posts: 1

Participants: 1

Read full topic

Jeff Atwood wrote:

Thanks to a generous grant from an awesome member of our community we now have the opportunity to hire someone on a contract basis to work on implementing a basic badge system in Discourse.

But first, we have to explain what the project is, and solicit feedback on it. Goals:

• A simple badge system, by that I mean implementable in a reasonable amount of time (6-8 weeks?).
• Something we are comfortable folding into core, and possibly enabling here and a few other willing partners, but will default to off for quite a while.
• Something with a solid base architecture that we can build on later to add more features.
• The contractor will work directly with the Discourse team, the project sponsor -- and you, the community!

So what should a simple, solid V1 badge system look like? The Discourse team discussed it and here's what we came up with:

What is a badge?

• Name
• Image
• Description
• Single/Multiple allowed
• (optional) ruleset for auto-award

I believe the images are very important here, because the value of a badge likes partly in the cool image (sticker, if you will) that you get associated with your account.

Staff should be able to create an unlimited number of manually awarded badges. The automatic badges will probably be hard-coded in V1 because we don't want to implement a complicated rules engine. Perhaps there could be some rough parameters, but that's it..

Where will badges appear?

I think a simple number underneath the username in the left gutter is best. The number indicates the number of badges you've earned. I thought about another glyph like a trophy or star, but I think letting a simple number send the message does it without all the weird emotional baggage of a glyph.

Clicking the number will take you to the badges page on the user's profile. The badge page should show:

• list of badges awarded aka "trophy case"
• when they were awarded
• by who, if manually awarded
• why, if automatically awarded
• how many times they were awarded (if multiples are possible)

How will badges be awarded?

1. We must support manual award of badges by staff members for any reason, to any user. This is essential.

2. We should also have a core of 10 badges that are auto-awarded for reasons that support our goal of civilized discourse. In other words, the badges encourage specific behaviors on the forum that help people have better conversations. These 10 should be exemplary, clearly excellent examples of good behaviors.

How will users be notified of the badge award?

Presumably through the notification stream as usual.

What should the 10 initial auto-awarded badges be?

And more importantly what positive community behavior do they encourage?

1. (name tba) Filled out your user profile completely including an avatar. Why? A community full of avatars and useful profile information makes it easier to see who you're talking to.

2. (name tba) Read the community guidelines (/faq) completely. Why? The community guidelines are everyone's recipe for civilized discourse.

3. (name tba) Cast a like. Why? Learning to give your fellow community members a hearty pat on the back -- and avoiding meaningless "+1" and "me too" posts --

4. (name tba) Quoted a post. Why? Quoting is crucial for flat discussions, learning to select text in another post (or use the quote button in the toolbar) helps keep discussions on topic and readable.

5. (name tba) Replied as new topic. Why? The best way to keep topics on topic is to learn when to take your discussion to a related topic rather than replying directly.

6. (name tba) Visited multiple days, read multiple topics. Why? Showing up regularly and reading a bunch of stuff the community is posting is the baseline of meaningful activity on any forum.

7. (name tba) Read a long topic, every post, all the way to the end. Why? Endurance! And yes @sam we will check that you didn't just mindlessly scroll as fast as you could to get to the end, either. Read Time Matters.

8. (name tba) Shared a topic that got clicked on. Why? Sharing topics with the outside world (using the share button and the URL that has your username on it) is how we attract new blood to any community.

9. (name tba) Flagged a topic or post. Why? We put a trashcan on every street corner so the city can stay clean. Everyone has to pitch in.

10. (name tba) Edited one of your posts. Why? We want people to be unafraid to edit their stuff to make corrections or improvements.

11. (name tba) Trust level promotion. Why? This is kind of a stealth way of getting education about what trust levels mean in here. You go up a level, you get directed to a page that explains your new "powers" in badge form..

I know, I know, you probably have a dozen badge ideas. But before you post,

• We want a limited list of badges in V1 to keep it simple. So the only way your badge is getting on this list is if it's way more awesome than one of the 10 that's already here.

• Before proposing anything, PLEASE consider first: what positive community behavior would your badge encourage? If you can't explain that first, don't do it.

Anything else?

I'm sure I've forgotten or neglected to cover a bunch of stuff. Feedback welcome. And remember, we're trying to keep this simple in V1!

Posts: 115

Participants: 27

Read full topic

Vikhyat Korrapati wrote:

It looks like the HTML sanitizer interprets < followed by any non-whitespace character as an HTML tag, causing the remainder of the line to be stripped out.

This causes problems when users enter smiley faces like <_< or <3. I think it would be a good idea to replace <s with < when the next character is not an alphabet before passing it to the sanitizer?

Posts: 4

Participants: 3

Read full topic

Ray wrote:

I started arranging categories in terms of color blocks, for example all red blocks are clumped together, blue blocks are clumped together, etcetera. But after tediously arranging them in personal sequence & saving it, it looks like it still got all jumbled up? Could there be a bit of entropy?

The only sequence that got saved properly starts with category "General Forum" all the way to "Issues & Warranty". Everything else below issues & warranty did not save properly in their preferred order.

It also looks like the category order does a bit of a shuffle dance once again when viewing it logged out.

http://idxclub.co/categories

Posts: 4

Participants: 3

Read full topic

Sam Saffron wrote:

I have been working on a heavily simplified plugin interface so its a lot easier for people to get started with extending Discourse.

For example:

Say you wanted to extend Discourse and add Ubuntu SSO to the dialog.

Place a file called plugin.rb in the directory plugins/ubuntu_sso

###plugins/ubuntu_sso/plugin.rb

# name: discourse-ubuntu
# about: ubuntu login support for Discourse
# version: 0.1
# authors: Sam Saffron, Marco Ceppi

auth_provider :title => 'with Ubuntu',
              :authenticator => Auth::OpenIdAuthenticator.new('ubuntu','https://login.ubuntu.com', trusted: true),
              :message => 'Authenticating with Ubuntu (make sure pop up blockers are not enbaled)',
              :frame_width => 1000,   # the frame size used for the pop up window, overrides default
              :frame_height => 800

register_css <<CSS

.btn-social.ubuntu {
  background: #dd4814;
}

.btn-social.ubuntu:before {
  font-family: Ubuntu;
  content: "U";
}

CSS

Run bundle exec rake assets:precompile
Restart you services.

Done, you have working Ubuntu SSO.

This interface is still not complete, as I finalize it I will be adding more examples here.

The /plugins directory is ignored by git so you can place your extensions there safely.

Longer term plans

1. Possibly use a Rails::Engine for greater flexibility
2. More hooks
3. Admin UI to enable or disable plugins (with multisite support)
4. Central plugin repo
5. External gem dependencies

If you are interested in extending the interface etc, let me know.

cc @marcoceppi

Posts: 93

Participants: 42

Read full topic

Claus Strasburger wrote:

I often have a Discourse instance open as a pinned tab (AKA application tab), to save some precious pixels. My browser (Chrome 31) rewards this with blinking the tab whenever something's new.
The problem: This event fires way too often, even if nothing changes. In particular, it fires if I switch to another tab. So my tab is left blinking constantly.

I found out Chrome does this whenever the page title changes. After some more searching, I found the page title is updated in Discourse every time the focus changes!
(see observes(..."hasFocus"...))

github.com

https://github.com/discourse/discourse/blob/7e984e369af424aa7478047da7e7d663b3b01685/app/assets/javascripts/discourse.js#L46

    var notifyCount = this.get('notifyCount');
    if (notifyCount > 0 && !Discourse.User.currentProp('dynamic_favicon')) {
      title = "(" + notifyCount + ") " + title;
    }
    // chrome bug workaround see: http://stackoverflow.com/questions/2952384/changing-the-window-title-when-focussing-the-window-doesnt-work-in-chrome
    window.setTimeout(function() {
      document.title = ".";
      document.title = title;
    }, 200);
  }.observes('title', 'hasFocus', 'notifyCount'),

  faviconChanged: function() {
    if(Discourse.User.currentProp('dynamic_favicon')) {
      new Favcount(Discourse.SiteSettings.favicon_url).set(
        this.get('notifyCount')
      );
    }
  }.observes('notifyCount'),

  // The classes of buttons to show on a post
  postButtons: function() {

Two questions:

• Is it really necessary to observe that property?
• If yes, can we add a check to only change the title property if it's different from the last one?

Posts: 9

Participants: 5

Read full topic

Neil wrote:

Continuing the discussion from Internationalization / localization, which has become a massive, unfocused topic:

Let's talk about translation tools and the process for getting translations done. If we were going to adopt a third-party tool to get translations done, which should be used? From the Internationalization / localization topic, there seem to be two leaders: Transifex and Localeapp.

Here are the things we need from a translation process:

• Notifications for translators when there's work to do, only for the languages they care about.
• No cost to translators.
• An easy way to upload our english yml files without clobbering anyone's work, ideally using an API.
• An easy way to get all the translations and add them to the Discourse repo in github. We must be able to automate this so it's a one-click operation!

I played with Transifex briefly, and it seems to have a very nice interface for translators. It has the ability to watch languages, leave comments on translations, tagging, etc.

Does anyone know about automation of this whole process?

Posts: 43

Participants: 15

Read full topic

FichteFoll wrote:

I'm not sure if this is a FireFox-only issue and I doubt it, but the procedure is as follows:

1. Open a page at a specific post (hotlink, permalink whatever), for example https://meta.discourse.org/t/welcome-to-meta-discourse-org/1/23
2. The post is vertically centered and highlighted with the (new) blue color.
3. Refresh the site.
4. The page now centers the previous post and highlights that. You can notice how the url changed to .../1/4.
5. Repeat until you reached the original post of the thread.

I can understand why this is happening and I also understand that it is hard to track which post I am actually reading, but in the described scenario this should not happen.

My use case that led to this discovery:
I opened a few posts that I wanted to read later in multiple tabs. I then wanted to reply to those but since I didn't have an account on the site yet I had to create one. Hence, I had to refresh all the tabs I opened and suddenly was wondering why I opened them in the first place because I could not remember why I would open the exact posts the view was focussed on.

Posts: 2

Participants: 2

Read full topic

Noah Crowley wrote:

We have implemented the SSO component into our ASP.NET MVC 4 website in order to allow users to automatically login to our Discourse forum directly from the website itself. As an additional feature, we'd like to add the capability to edit their Discourse preferences - primarily the email and subscription settings - from a Profile Settings page on our website. How would I go about editing preferences like this using the API? I'm a C# programmer, new to Ruby. I understand JSON, but I'm not quite sure how to properly send the data required to do this to the Ruby that Discourse runs on.

All help is very much so appreciated. Thank you!

Posts: 2

Participants: 2

Read full topic

Sebastian wrote:

Has the founders of DC software intended to for it be more of a quora/stackoverflow like forum where the main intention seems to be exchange of information/solve problems?

It feels like some groups like coders prefer that model, whereas a group of cat lovers might prefer the chit chat mode more. Not sure what a chit chat mode like forum looks like to be honest. But the view top answers and summarize thread definitely give a more pragmatic, male feeling of the forum.

Posts: 3

Participants: 2

Read full topic

Jeff Atwood wrote:

So you'd like to set up reply via email support for your Discourse forum.

Good news! Unlike outgoing email, which is completely unsuitable for typical consumer mail services like GMail, Outlook, Yahoo Mail, etc, incoming reply via-email tends to be extremely low volume. Unless you have a truly massive forum with a zillion people replying to notification emails, you are probably safe using, say, Gmail.

So here's how to set it up using GMail as an example.

1. Create a new GMail account. Assign it a username like discourse-replies@example.com or in my case, replies@example.com. Give it a strong password.

2. Log in to this account via http://mail.google.com. Accept the terms of service, enter the CAPTCHA, load the default web interface. Maybe send a test email, receive a test email. Kick the tires. POP3s access is on by default in Gmail, and all incoming emails will be retained -- but you can tweak as needed in Settings, Forwarding and POP/IMAP.

3. Visit the Discourse settings and check the Email tab.

4. Change the following settings:

   • reply_by_email_address enter replies+%{reply_key}@example.com
   • pop3s_polling_username enter replies@example.com
   • pop3s_polling_password enter the password of that email account
   • pop3s_polling_host set to pop.gmail.com
   • pop3s_polling_enabled set to true
   • reply_by_email_enabled set to true

Pasted image673x535 43.4 KB

There is a certain amount of polling that goes on for this account, but we've set it to a low, safe volume.

We do assume that your email service supports plus addressing aka address tags, so check that before starting. GMail obviously does.

If everything is working, you should see the footer of all notification mails now tell you that you can reply either by visiting the website, or replying via email:

To respond, reply to this email or visit {topic URL} in your browser.

Posts: 1

Participants: 1

Read full topic