Continuously deploying meta.discourse.org

July 22, 2014, 7:00 pm

≫ Next: Allow badges in category security settings

≪ Previous: You should not be allowed to change Lounge security

$

0

0

Sam Saffron wrote:

I just flicked the switch that continuously deploys meta every time we have a successful build (which triggers a merge of tests-passed branch)

There are quite a few reasons for this.

1. Our default docker templates now target the "tests-passed" branch, I want meta to be a place where people can go to always confidently see what it looks like.

2. This keeps us honest as developers, in the past I blogged about the importance of smoke tests, if for any reason we break master we now have a very clear public warning sign. We will stop everything and fix master and ensure an automated test will stop incidents like this in future.

3. I really want meta to go through every release of Discourse, that ensures we don't have boobytrap migrations and other weird upgrade conditions that are untested.

I really like doing this and think it will lead to a more stable master branch in the long term.

This new system is in effect now

Posts: 4

Participants: 3

Read full topic

---

Allow badges in category security settings

July 22, 2014, 10:45 pm

≫ Next: Upgrading from a release way in the past

≪ Previous: Continuously deploying meta.discourse.org

$

0

0

Tuan Anh Tran wrote:

Now that we have trigger for badges, should discourse allow badges to be used in category security setting along side with groups?

Posts: 7

Participants: 2

Read full topic

Upgrading from a release way in the past

July 22, 2014, 11:44 pm

≫ Next: Auto-set profile picture when signing up using Facebook

≪ Previous: Allow badges in category security settings

$

0

0

Matt Ginzton wrote:

I set up a private instance of Discourse a while ago, for a project that really only matters once a year, so it got ignored since October 2013.

I went to freshen up the Discourse install so people can use it again this year and find

• Discourse recommends I upgrade, of course: from 0.9.6.4 to current 0.9.9.13
• in the meantime, you've rolled out Docker and deprecated the manual install method I previously used
• following the upgrade steps for the old method (which I have successfully followed before), I get a fatal error about database "discourse" does not exist.

Specifically, when doing the rake db:migrate step (or the succeeding rake assets:precompile step), I get ActiveRecord::NoDatabaseError: FATAL: database "discourse" does not exist. I tried just starting up the new 0.9.9.13 installation anyway and unsurprisingly it doesn't start; I rolled back to the backup of the previous version.

(See https://gist.github.com/metamatt/ca83bc4c9afa83b9220c for the full error message and backtrace.)

I'm wondering if some of the databases or tables have been renamed over time, and the upgrade scripts know how to do this upgrade piecewise (so people who were following along closely with most of the interim releases didn't have problems) but not all at once; do I need to upgrade to one or more interim versions?

Or is there manual DB surgery possible to make this work?

I'm expecting part of the answer involves moving over to the Docker based install which I'm happy to do, but the instructions there start with "upgrade to the absolute latest version of Discourse" before attempting the transition to Docker, so.

Or if there are snapshots of old known-good Docker-based versions and the Docker-based upgrades are more reliable, maybe I could upgrade the old way to the same version as one of these snapshots, then move to Docker, then upgrade to the present?

I would love to get on the current version (0.9.9.13) and packaging method (Docker) without losing the real user posts that are already in my existing DB. Thanks for any help.

Posts: 1

Participants: 1

Read full topic

Auto-set profile picture when signing up using Facebook

July 23, 2014, 12:36 am

≫ Next: Popups are broken

≪ Previous: Upgrading from a release way in the past

$

0

0

Joseph Lormand wrote:

Continuing the discussion from Configuring Facebook login for Discourse:

I think this should be a built in functionality, available as an option in settings. Having a smooth transition into the forum makes a big difference when someone has just signed up, and having their profile picture grabbed from Facebook seems much better than using the current default letter box image.

Posts: 1

Participants: 1

Read full topic

Popups are broken

July 23, 2014, 2:44 am

≫ Next: Numbered/Bulleted List breaks undo/ctrl-z data

≪ Previous: Auto-set profile picture when signing up using Facebook

$

0

0

resure wrote:

So, I click "Sign In" and nothing happens (almost): http://i.imgur.com/U860LQ9.png -> http://i.imgur.com/fl2SWx5.png. Looks like another popups (like file upload) are broken too . Try.discourse.org is fine, but meta and all other up-to-date instances are affected. Tried on Firefox, Chrome, Safari, Safari on iOS and Chrome on Android.

Posts: 5

Participants: 3

Read full topic

Numbered/Bulleted List breaks undo/ctrl-z data

July 23, 2014, 4:15 am

≫ Next: How I embed Discourse on website?

≪ Previous: Popups are broken

$

0

0

Joseph Lormand wrote:

Typing out a large post can take some time, so it's always nice to have undo functionality in the background in the event an error is made. I ran across what I think is a bug when using numbered lists, where using the button to create one seems to greatly damage or sometimes completely discard all change history for purposes of undo. Upon a bit of tinkering it looks as though the bulleted lists have the same problem. In fact, this is my second time writing this post because I tested a bulleted list on it just to see and the process of trying to undo that and redo back to a partial salvageable version created such a mess I just started over. Looks like most of the buttons above the text box interact with undo in some strange way, but the lists seem to be the most severe.

Posts: 1

Participants: 1

Read full topic

How I embed Discourse on website?

July 23, 2014, 4:44 am

≫ Next: Precompiled assets across multiple production servers

≪ Previous: Numbered/Bulleted List breaks undo/ctrl-z data

$

0

0

Includektt wrote:

I want to embed Discourse on community website.
For example:http://discourse.ubuntu.com/latest
:http://discuss.howtogeek.com/

But i don't know Where should I edit.

My english is poor...
So please tell me easy english

Posts: 2

Participants: 3

Read full topic

Precompiled assets across multiple production servers

July 23, 2014, 5:18 am

≫ Next: ./launcher rebuild not creating /var/nginx/ folder

≪ Previous: How I embed Discourse on website?

$

0

0

tim wrote:

Question about precompiled assets on a multi-server production environment. The sprockets require_tree instruction does not define any ordering of its inclusions. On our two servers running identical code and identical versions of gems etc (same AWS AMI), the precompiled asset fingerprint is different due to the undefined nature of file system ordering.

I'm not a rails person normally so I have no intuition as to how people get around this problem normally. The require_directory sprockets command includes in lexicographical order, but does not recurse.

How is one meant to get around this problem for Discourse? Or is this not a problem and I have just misconfigured something incorrectly.

TIA!

Posts: 3

Participants: 2

Read full topic

---

./launcher rebuild not creating /var/nginx/ folder

July 23, 2014, 5:50 am

≫ Next: Topics appear in Unread() well before 'n new posts' banner

≪ Previous: Precompiled assets across multiple production servers

$

0

0

Jeffrey C Witt wrote:

After a rebuild of my web_only container, my site does not immediately work.

./launcher logs web shows the following repeated errors:

 nginx: [emerg] mkdir() "/var/nginx/cache" failed (2: No such file or directory)

I have been able to solve this problem by manually SSH into the web container and creating the nginx folder like so mkdir /var/nginx. After that, everything seems to work fine.

Any ideas ways the /var/nginx directory is not being created?

Posts: 4

Participants: 2

Read full topic

Topics appear in Unread() well before 'n new posts' banner

July 23, 2014, 6:05 am

≫ Next: Default gzip settings?

≪ Previous: ./launcher rebuild not creating /var/nginx/ folder

$

0

0

probus wrote:

It seems like Unread() is a lot quicker and more reliable to react to new posts than the banner in topic list. Very often I see the numbers changed in Unread() when someone posts into one of the topics I'm tracking with no notification in the topic list. Sometimes they both appear, but the Unread() count is correct, while the 'n new posts' is off. For example Unread(4) and only '2 new posts' in the banner.

Why is there a difference? How often are these updated anyway?

Posts: 1

Participants: 1

Read full topic

Default gzip settings?

July 23, 2014, 6:16 am

≫ Next: Unread/new badge style?

≪ Previous: Topics appear in Unread() well before 'n new posts' banner

$

0

0

Brad Huber wrote:

So, like an idiot, the random tutorial I followed (not one of the official ones) never mentioned adding a root password or ssh key to the app.yml file, so I cant ssh into my container. I just want to check the gzip settings in the nginx config, and dont want to shut down the forum, destroy the container, and bootstrap it again unless I need to actually go in there and change anything.

(I'll need to do it eventually, I know, but I'd rather do that late at night, and I need to know the gzip settings today)

Posts: 6

Participants: 4

Read full topic

Unread/new badge style?

July 23, 2014, 7:49 am

≫ Next: Create topics notice remains visible

≪ Previous: Default gzip settings?

$

0

0

kris wrote:

The style of the new/unread badges is feeling a bit off to me.

Maybe that's a good thing because they stand out more? are they too similar to categories without the border radius?

Screenshot 2014-07-23 10.35.23.png1295x380 102 KB

Screenshot 2014-07-23 10.30.03.png1170x316 76.5 KB

Posts: 8

Participants: 5

Read full topic

Create topics notice remains visible

July 23, 2014, 8:08 am

≫ Next: Discourse site can't be crawled as robots.txt fetch failed

≪ Previous: Unread/new badge style?

$

0

0

Matthieu Wipliez wrote:

The "create topics notice" that mentions that as an admin I have to create 5 public topics and 30 public posts is not going away, even though the forum has now exceeded these thresholds. I haven't changed most options, since the defaults are good for me (I'm running a public forum). I am happily using the latest version, and that is the only minor annoyance I have encountered when using Discourse, good work guys

Posts: 1

Participants: 1

Read full topic

Discourse site can't be crawled as robots.txt fetch failed

July 23, 2014, 9:12 am

≫ Next: Oneboxing oneboxed post lacks context

≪ Previous: Create topics notice remains visible

$

0

0

node wrote:

I can't add my website to google through webmaster tools as it cannot access robots.txt. My site is only accessible through ssl, and I've tried going to https://example.com/robots.txt and can see my robots.txt file but bots can't crawl it for some reason.

I've tried using the Yandex tool which works for https://meta.discourse.org/robots.txt but returns "Robots.txt could not be loaded." for my site.

On webmaster tools I get robots.txt fetch failed, but when I click "see live robots.txt" I can see my robots.txt file. I've tried this on different devices and different ISPs without being logged into discourse.

Posts: 3

Participants: 2

Read full topic

Oneboxing oneboxed post lacks context

July 23, 2014, 9:53 am

≫ Next: Very Inconsistent between preview pane and post

≪ Previous: Discourse site can't be crawled as robots.txt fetch failed

$

0

0

Arpit Jalan wrote:

As described in title, here's the repro:

Posts: 7

Participants: 4

Read full topic

---

Very Inconsistent between preview pane and post

July 23, 2014, 9:57 am

≫ Next: Do we need to have media=screen?

≪ Previous: Oneboxing oneboxed post lacks context

$

0

0

Matches wrote:

See below for example from TDWTF - Compare where the strike tag is, and the resulting actual post strike through. The preview pane shows a strike through all but the last sentence.

Continuing the discussion from Burn the witch!:

@PJH This is actually a bug, because you can't like topics that are closed by moderators. You can only like topics auto closed by the system.
[EDIT]

Correction, it appears to be mobile specific. Investigating.

Nevermind. Apparently I'm just wrong in total.

Posts: 4

Participants: 2

Read full topic

Do we need to have media=screen?

July 23, 2014, 10:13 am

≫ Next: Migrate db from old Discourse to new without updating

≪ Previous: Very Inconsistent between preview pane and post

$

0

0

Thomas Smyth wrote:

Is there any reason the main stylesheet link tag has to have media="screen" explicitly? In my plugin, I have a simple page that I want to print and I've added my own media query to hide the stuff I want to hide, etc., but because the main link tag has media="screen" then all my styles are gone so the page looks like crap and my media queries are gone too. If I remove media="screen" using the inspector then it works fine. Thoughts?

Posts: 1

Participants: 1

Read full topic

Migrate db from old Discourse to new without updating

July 23, 2014, 10:28 am

≫ Next: Tagger not working on multisite install

≪ Previous: Do we need to have media=screen?

$

0

0

Sedrickcz wrote:

Hi,

I have instance of Discourse without db backup (0.9.8.3) and I need backup my db before updating to new version and I have to be sure, that I have fully functional db backup that works on the newest version.

I tried this: sudo -iu postgres pg_dump --no-owner --clean discourse | gzip -c > /shared/today-backup.tar.gz and upload it to new version, but it doesn't restore it

Any advice how to do it? Thank you.

Posts: 8

Participants: 5

Read full topic

Tagger not working on multisite install

July 23, 2014, 10:52 am

≫ Next: Mismatched [quote]-tags are handled incorrectly

≪ Previous: Migrate db from old Discourse to new without updating

$

0

0

Jeffrey C Witt wrote:

@lightyear @riking Can you guys think of any reasons why the tagger plugin isn't working with a multisite install?

Right now I don't get any errors (at least I don't think) when rebuilding the web_only container, but when I got to the admin->tags on the second_site, I get an error. And generally on both sites the the plugin doesn't seem to be functioning correctly.

Currently, I have removed it -- I'll have to rebuild the container to reproduce the error. At present, I was just wondering if there is some a priori reason why the plugin would not work with a multisite install.

If not, I'll try to reproduce the error and provide you with more info.

Posts: 7

Participants: 2

Read full topic

Mismatched [quote]-tags are handled incorrectly

July 23, 2014, 11:44 am

≫ Next: Ability to disallow changing passwords

≪ Previous: Tagger not working on multisite install

$

0

0

Matches wrote:

This is the first post

Posts: 10

Participants: 3

Read full topic