The current bbcode URL handler is broken and will send invalid JSON-ML arrays up the call stack if the contents of a [url]...[ /url] (space added to avoid triggering the bug) tag are not actually recognized as a URL; more specifically, the bug pops up unless the autolink dialect has already transformed the tag content into valid JSON-ML representing a single a tag...
I've been using Discourse for a few months (thank you!), originally using a Bitnami stack. Yesterday I bit the bullet and got the latest version running in Docker (on a Linode server, the same one the old Bitnami stack was running on). I followed the migration instructions from https://meta.discourse.org/t/manually-create-and-restore-discourse-backups/18273, and almost all the content and data came through.
The one thing that isn't working is the avatar thumbnails. Locally-uploaded avatars only show up at certain sizes, for example the 50px one doesn't show up (http://community.dinopoloclub.com/user_avatar/community.dinopoloclub.com/straan/50/221.png) but that user's profile page looks fine (http://community.dinopoloclub.com/users/straan/activity). I reuploaded my avatar and, after changing the filename to force it to regenerate the thumbnails, it works everywhere. So newly uploaded avatars are ok. Gravatars aren't working at all.
I've double-checked and all the uploads came across from the previous install. The directory /srv/discourse/apps/discourse/htdocs/public/uploads/default in the old Bitnami install was moved to /var/www/discourse/public/uploads/default in the new Docker container. I've run rake avatars:clean and avatars:refresh to no avail.
I am making the call and officially dropping support for 1.9.
The final catalyst here is that we are stuck and can not upgrade Sidekiq until we make this call. I do not want us to ship with software that contains potential security holes, we must embrace the future.
For those on our Docker setups this news really means nothing, you have been on Ruby 2 for ages. For developers, be sure to install Ruby 2 or better still 2.1
Following the install guide (https://github.com/discourse/discourse/blob/master/docs/INSTALL-digital-ocean.md) I successfully installed the latest stable version on my server. It worked well and I registered as administrator. But after I rebooted the server, the docker container did not start automatically. when I use ./launcher start app, it shows:
cid found, ensuring container is started Error response from daemon: Cannot start container [container id]: (exit status 1) 2014/09/25 13:28:43 Error: failed to start one or more containers.
When I tried to rebuild, it told me that "github.com" cannot be resolved, but I am sure that I can connect to github.com (wget, ping, etc are all okay) The only solution seems to be purge the lxc-docker package, install that again, and reinstall discourse completely. By this way, however, all my data are gone. I can install discourse again normally, but after I reboot the server again, it went into the same error again. Is there a solution for this problem?
An new invited user contacted me because he couldn't delete own its account. FYI, he posted 0 messages. I tried as admin and I can't too and get the following message:
Il y a eu une erreur lors de la suppression de l'utilisateur. Veuillez vous assurez que tous ses messages ont bien été supprimmés avant d'essayer de supprimer l'utilisateur.
When I start to type a colon : it appears a popup with the available emoji. Then I type the open parenthesis ) to make the smile, but it shows the frowning icon, here's a screenshot:
If I select the proposed icon, the text is :frowning: and in the preview I see the frowning icon
I saw that there is a build mechanism for Debian and the installation works quite well for me. Why is this not mentioned in the setup guides with a link to enter link description here?
Since I logged in here a few minutes ago, I've been unable to access any "suggested topic" by using j/k to navigate to it and pressing "Enter" (my usual method of navigation). Every time, I get a message "Sorry, we are unable to load that topic, possibly due to a connection problem" or something similar and it reloads the topic I was viewing.
The links work if visited by mouse - and seem to work by keyboard after being visited by mouse, although I'm not entirely sure about that.
The problem does not occur on our own instance; only on Meta.
Admittedly two changes have occurred to the site and I'm not sure which could be causing this issue. We recently upgraded our forum to 1.1.0beta3, and everything seemed great. Then last night we installed an SSL certificate on our server, enabled https in the Discourse admin, and again, everything seemed great until I started testing for mobile.
I didn't test for mobile after the beta upgrade so I'm not sure which event triggered the issue.
The header area loads on mobile, and then nothing else. I've tested using BrowserStack and a series of other mobile devices (Android Tablet, iPhone, iPad). The browser "hangs" like it wants to load the rest of the topics but it never does.
The orange is dead space on Latest (and New) topic list. You can click the bottom row or the top row to get to the topic, but there is a bit of a gap in the middle.
We just started using discourse inside of an iframe on our site. When a topic gets more than 3 or 4 posts and a user scrolls down loading... is shown and never goes away. I can provide screen shots or any other info that may be useful. Are there any known issues with iframes?
I manage to keep Discourse translation for Chinese shining recently. But it's super annoying to keep the translation up to date when you upgrade to the new version.
I can't promise the translation on stable is stable. It hurts. The problem is because Transifex is catching with master. But it's not easy to put them back on the old branch since the old string may be deleted on Transifex.
Manually updating translation is a work around but it's not quite right. Any thoughts? Does Transifex provide anything for that? Or build the script to deal with that?