@terraboss wrote:
My admin panel said, I've read 394 posts (?) but the public user list is telling me 198. Which value is the right one?
Why is this discrepancy?
Posts: 3
Participants: 2
@terraboss wrote:
My admin panel said, I've read 394 posts (?) but the public user list is telling me 198. Which value is the right one?
Why is this discrepancy?
Posts: 3
Participants: 2
@flameens wrote:
Hi there, I've had some issues importing the full names from a mybb installation and I want to delete all the users full names so they can correct them later from their own preferences. I want to run the following command on the rails console:
User.all.each do |u| u.name = "" u.save end
Is this the correct syntax?
Posts: 4
Participants: 2
@tobiaseigen wrote:
We're looking for a developer to write up a scheduled task that runs daily to enforce the following user preferences, based on group. Ideally this would be contributed as an admin-customizable plugin to the community so it can also be used by others (like people involved in this topic) that might also want this functionality.
However we do have an immediate need for this functionality so we are open to quick and dirty solutions, including help with postgres queries we can run manually to enforce user preferences. Open to proposals - please PM me.
For our specific purposes, the requirements are as follows.
If a user is in the ourorg_staff group, the following user settings are enforced:
- default email private messages - SELECTED
- default email direct - SELECTED
- default email always - SELECTED
- default categories watching - ANNOUNCEMENTS
- default categories tracked and muted - NOT ANNOUNCEMENTS
If a user is in the ourorg_leadership group, tthe following user settings are enforced:
- default categories watching - LEADERSHIP
- default categories tracked and muted - NOT LEADERSHIP
Other settings, especially other category notification settings, are not affected.
Posts: 1
Participants: 1
@erlend_sh wrote:
A bug report means something is broken, preventing normal/typical use of Discourse.
How to report a bug in Discourse
Make sure the bug isn’t already resolved. Search for similar issues in the bug category.
Make sure you can reproduce your problem on the default sandbox at try.discourse.org
- If you’re also testing on your own Discourse development instance, be sure to use the Master release branch, aka the Tests Passed channel.
- If possible, submit a Pull Request with a failing test, or;
- if you'd rather take matters into your own hands, try fix the bug yourself
(you need to read up on the Discourse Development Contribution Guidelines for these last two.).Make a report of everything you know about the bug so far by opening a topic on meta about it.
When the bug is fixed, you can usually expect to see an update posted on the reporting topic.
Posts: 1
Participants: 1
@swisspol wrote:
For instance
- <kbd>hello</kbd> - <kbd>world</kbd>
The bubbles are touching each other:
Posts: 1
Participants: 1
@bdl wrote:
I've used three VPS providers primarily and they are, by far, the three largest in the U.S.: Amazon EC2, Digitalocean, and VULTR. They are all on what I consider top networks with dozens of backbone peers and the best that anyone can do for a budget VPS (i.e., not paying up for a CDN like Fastly yet). I believe that each guarantees 100% uptime in their SLA.
Anyway, I found a great site for testing the network speeds of these hosts to various points in North America. I'd always assumed that of the major points of presence in the U.S., probably Chicago or Atlanta are the ideal places to locate web servers that are not on a CDN. Being at a major Internet crossroads more in the middle of the country (and Georgia may be the "east coast," but Atlanta is geographically as far west as it is south of the Northeast) ensures that you're never too far from any particular web surfer. Armed with these new statistics from CloudHarmony, I now see that I was right.
In short, here are the benchmarked average latencies to North American Internet users. Moral of the story: VULTR is the most bang for your buck, but location matters more than provider. Host in Chicago or Atlanta if you can; avoid hosting in the west (as the North American population still skews heavily toward the east), and especially avoid locations at the extreme tips of the country (such as Miami and Seattle).
I had already chosen VULTR/Chicago before seeing these benchmarks, and now I feel vindicated! It's the only VPS location to come in at under 100 ms on average to typical North American Internet users.
VULTR (Chicago), 99 ms
VULTR (Atlanta), 101 ms
Amazon EC2 (Virginia/DC), 105 ms
VULTR (Dallas), 106 ms
VULTR (New York), 106 ms
DO (New York 1), 109 ms
DO (New York 2), 109 ms
DO (New York 3), 110 ms
VULTR (Los Angeles), 119 ms
VULTR (Miami), 121 ms
VULTR (Silicon Valley), 124 ms
Amazon EC2 (Silicon Valley), 125 ms
DO (San Francisco), 130 ms
VULTR (Seattle), 133 ms
Amazon EC2 (Portland), 144 msFor those of you wondering how much of a difference a CDN could make, as a rule they obviously blow even VULTR/Chicago out of the water.
SoftLayer, 41 ms
Akamai, 48 ms
CloudFlare, 48 ms
Fastly, 59 ms
Amazon CloudFront, 75 msAlthough, as you can see, some more than others. Amazon CloudFront is more similar to VULTR's single Chicago location than it is to SoftLayer, CloudFlare, or Akamai here.
Posts: 4
Participants: 3
@David_Moffett wrote:
Hey guys, I spent some time doing some searching and couldn't find an answer to the following question:
With the poll plugin is there any way to enable the display of which users have selected which options? Specifically I'm working on a new project over at my community site and I'm trying to figure out which users are interested in helping with various aspects of a project. As an example let's say my poll looked like this:
[poll name=interested type=multiple] - Brainstorm, discuss, and critique ideas; basically just offer creative advice - Crunch numbers and perform critical analysis - Write back end code - Write front end code - Testing - Not interested in helping at all [/poll]
Each user may select one or more options. I'm interested in knowing which users select which options. Is there an attribute or anything to let me do this?
Thanks!
Posts: 3
Participants: 2
@gerhard wrote:
I guess this is a low priority bug...
Steps to reproduce:
- Visit the sorted bug list by entering https://meta.discourse.org/category/bug?order=op_likes&status=open in the browser's address bar
- Click on the Discourse logo and navigate to the Latest page
The "Likes" column stays visible, but it shouldn't.
Posts: 1
Participants: 1
@sam wrote:
Letter avatars are our 4th most expensive route globally.
This route is responsible for retrieving and generating unique letter avatars.
For scale of the enormity of the current cost, in the last 24 hours we spent 30 minutes of CPU time in the last 24 hours on meta alone, generating these avatars. At our hosting scale it's as though we have single CPU dedicated permanently for letter avatar generation.
CPU alone is not the issue, we also have all these pesky web requests to deliver these little images to users.
Instead we would like to investigate another approach here that is more efficient.
Option 1: ship a font and use CSS
The simplest and probably most efficient way would be to pick a pretty font from: https://www.google.com/fonts and use CSS to render this stuff.
If we were to do this with "font + CSS" we must use a consistent font, otherwise positioning will be hellish and the UX. Centering the letter is not going to be super easy but probably doable.
Pros:
- Simple to build
- Will be very fast regardless of avatar count on screen
- Embed freindly
Cons:
- Centering in CSS is hard
- We need to ship an extra icon font
- CSS for this will be quite long
- Not RSS friendly (unlikely to be even doable in a sane way)
- Not Email friendly (will require a fair amount of work to get it to work in email)
Suggested markup:
<div width='100' height='100' class='letter-avatar letter-color-1 background-color-76'>A</div>
Option 2: use a font and Canvas
A slightly more complex option would be to use a font + canvas api to generate an image. This will make it slightly easier to center the letter. It may be slightly easier to integrate as well.
Pros:
- Centering in canvas is easy
Cons:
- We need to ship an icon font
- We need to run a "post processor" in JS on all letter avatars
- Impossible to do in Email
Option 3: use SVG
We could consider using a 100% SVG solution, ship all the glyphs in SVG. If we were to follow this we would have to have to ship an extra asset, cause I would not want to bloat vendor.js or application.js
Pros:
- Can be extremely easy on embedding if all SVG is inlined
Cons:
- Requires and extra asset
- Very complicated to build
- Will bloat cooked Markdown for quotes if we inline SVG
- Unlikely to be doable in Email
Option 4: Free letter avatar service
Have Discourse host a single, "free", letter avatar service that any Discourse customer can use, wire it in as default.
Pros:
- Simple to integrate into Discourse
- Can be default on with a simple off switch
- Allows us to provide custom "default" avatar styles
- Can run on Digital Ocean + Cloudflare and have very low running cost.
- Simple to build
- We can upgrade the avatar algorithm at will
- Everyone gets to use 1 cache
- Can reuse existing code
Cons:
- We need to run yet another service
- We still got to maintain the fallback
- If cloudflare has issues, no letter avatars (need to check on great firewall of China)
- Some people may have privacy concerns (but they still have an off switch)
Suggested markup:
Options 1, super private
<img src="https://avatars.discourse.org/a/40841852d1bb4460907de74de8b0cfba?width=50&height=50&v=1">
Guid is generated using username + random hidden site setting hash
Option 2, high letter avatar reuse and cache use
<img src="https://avatars.discourse.org/sam_saffron?width=50&height=50&v=1">
Option 3, super high cache reuse, good privacy
<img src="https://avatars.discourse.org/letter/s/100_150_10/50.png&v=1">
(100, 150, 10 are the colors being used)
(I lean toward option 3 here)
The one thing that is clear is that something needs to be done.
After thinking this through I think #4 are the best approach we can take, it leaves people a safety valve and is the least intrusive change code wise.
Also going forward it allows us to provide more "default avatar" styles that people can choose from which can be handy.
Further work
#3 route cost wise is the standard user avatar route, this happens because we are resizing images on demand (we do keep a cache, but still). We must investigate if there is a more efficient way to size down these images, if the original avatar is already optimized, we may be able to skip the optimization stage altogether. We may also consider adding rmagick so we do not spawn a process each time we resize an avatar.
Posts: 30
Participants: 11
@Jong wrote:
When I attache a url, I got a error on console.
How can I fix this? What's the problem?
Thanks for helping.
[Error] Failed to load resource: the server responded with a status of 404 (Not Found) (onebox, line 0) http://tidev.kr/onebox?url=http%3A%2F%2Fwww.slideshare.net%2Fdreamcamp%2F2015-8-stayes%3Frelated%3D1&refresh=false
Posts: 3
Participants: 2
@tobiaseigen wrote:
This.
We're willing to pay to make it happen sooner rather than later. PM me if interested.
Continuing the discussion from Add Tags notification settings on user preferences:
Posts: 1
Participants: 1
@erlend_sh wrote:
Piwik is an open source analytics platform. Users can choose between self-hosting or paying for professional hosting in the cloud.
To enable Piwik analytics on your Discourse forum, all you have to do is insert your tracking code in the right place.
Go to
Admin / Customize / CSS/HTML / </body>
and insert:<!-- Piwik --> <script type="text/javascript"> var _paq = _paq || []; _paq.push(['trackPageView']); _paq.push(['enableLinkTracking']); (function() { var u="//<!-- URL HERE -->"; _paq.push(['setTrackerUrl', u+'piwik.php']); _paq.push(['setSiteId', <!-- WEBSITE ID HERE -->]); var d=document, g=d.createElement('script'), s=d.getElementsByTagName('script')[0]; g.type='text/javascript'; g.async=true; g.defer=true; g.src=u+'piwik.js'; s.parentNode.insertBefore(g,s); })(); Discourse.PageTracker.current().on("change", function(url) { _paq.push(["setCustomUrl", url]); _paq.push(["setDocumentTitle", document.title]); _paq.push(["trackPageView"]); }); </script> <!-- End Piwik Code -->
Be sure to change these two:
var u="//<!-- URL HERE -->";
_paq.push(['setSiteId', <!-- WEBSITE ID HERE -->]);
That's it!
Posts: 1
Participants: 1
@dmitry_fedyuk wrote:
As we agreed with Jeff Atwood all the Discourse.PRO plugins are GPL now.
Jeff said I can trust that the Discourse Meta community will not share my software without paying for it, so I will try to trust.
Please note that I set license price as 5% of my development costs. For example, if a license price is $100 than I invested $2000 in development (estimated in my work time).
If you estimate my work time and development costs lower please do not buy my software and I will not sell it to you.
Please note that if you buy my software and share it you just steal my salary and lead me to bankruptcy. I have no office job and my software is my only revenue.
Please note I am not seeking for any job or projects or donations. I want to do my product business only: do my software and sell my software to the mass market. I do custom development only as a bonus for VIP clients: https://discourse.pro/plans
Posts: 6
Participants: 5
@adwiv wrote:
Currently, we have to upgrade each repository separately.
So if we have a few plugins, we have to click upgrade on first repository, wait for it to pull, compile and restart and then click on the next upgrade and so on. This becomes more cumbersome as the number of plugins increase.
Would it be possible / make sense to have a single click "Upgrade All" button? We generally want the latest versions of all plugins, so it makes sense to me to upgrade all of them together.
Also, a single "Upgrade All" may even prevent rare conflicts between older version of plugin and newer discourse, during the period of upgrade.
Posts: 1
Participants: 1
@Fabio_Machado_de_Oli wrote:
Continuing the discussion from Google's Mobile Friendly Test:
Posts: 1
Participants: 1
@pjv wrote:
Help; I just upgraded my forum from 1.3.4 to 1.3.5 and it seems to have killed it.
I'm using the docker install.
I tried going to
/var/discourse
and doinggit pull
and./launcher rebuild app
and it's the same. I can see my custom header but underneath that is just white nothing.Any suggestions on how I can get my forum back online? (I do have daily backups - though I cannot reach the backups interface).
Posts: 12
Participants: 4
@BCHK wrote:
I just upgraded to the latest version of software last night - but now many of my users are complaining that some of the threads no longer have a reply button on them, so they can't reply. Its not on all topics, just some it seems. Anyone else seeing this problem?
Posts: 16
Participants: 7
@johnmuhl wrote:
Continuing the discussion from Feature request: documentation for poll feature, and other related features:
The linked article "does not exist or is private". Is there a new location for polls documentation?
Posts: 3
Participants: 2
@Lee_Ars wrote:
After doing a
launcher app rebuild
this morning to get onto the latest Discourse, I've run into issues with loading some customized site assets that worked prior to the upgrade: self-hosted@fontface
assets, and, weirdly enough, my site's favicon.I'd created a directory at
/var/discourse/shared/uploads/default/customizations
to hold this stuff, along with the site's big and small header images. Looks like this:As of this morning, most, but not all, of the things in that directory are showing 404. For example, in this screenshot, we've got the file
d-logo-wide-copse.png
, which is the site header image, and it's definitely being loaded. Thewoff
files aren't:I can also successfully grab the site headers out of that directory with curl, while the other files (including the favicon!) return Discourse's 404 error.
As a workaround, I tried uploading the favicon to a temporary topic in Staff, and that appeared to work; I can do this for the
@fontface
files, but man, that's an ugly, ugly, ugly ass workaround. Do you guys have any insight into why the neat and tidy custom location is suddenly puking?(To forestall a response I've seen elsewhere, no, using Google web fonts or another offsite font-host doesn't give me the flexibility I want—been there, done it, abandoned it.)
Posts: 4
Participants: 3