Back in 2019 I experimented with Brotli HTTPS compression - an option that should improve the site’s speed. It turned out Cloudflare had some bugs in their implementation that messed with Discourse.
I went through their documentation again and Discourse even recommends enabling it now, so let’s try again!
Update: I disabled this again. While it apparently did speed up the site for some people, it also broke most of the real-time interactive features (updating post lists, notifications etc). These are more important to a community than the speed bump. Cloudflare is aware of the issue and I’ll monitor its resolution.
Can you tell a difference in loading/browsing speed? Leave a response in the poll. If run in to any issues, please leave a comment.
I was playing with Brotli too, and it really speeds up websites’ performance in comparison with Gzip.
Also, you can try out and check OpenLiteSpeed solutions. It works faster than both Apache and Nginx servers, but I don’t know whether it’s applicable for Discourse or not. It works well with the WordPress though.
I would be reluctant to trade stability for “speed.” Particularly not unless everyday site users are regularly complaining about “speed.” As a card-carrying geek, I say: “don’t borrow trouble.”
Oddly enough, I just experienced a “spinning circle until reload” on this very page, while the others loaded rapidly. However, I am sure that this was a one-time occurrence: as you see, the problem went away.
As for myself, I have never observed “a loading-speed problem” here, and so do not detect an actionable issue. I cannot speak for others, nor do I see your admin/owner technical perspective. (But I do understand it.)
The speed wasn’t a problem, but if we can achieve a faster site at no cost that’s something to be explored. You’re the one suggesting there’s a problem, not me
I technically-understand and respect your comment – but, I never suggested “a problem.” (Quite the opposite: I don’t see one.) However, yes, if you can get something equally-stable for free, “feel free!”
Well, you wrote I was trading stability for speed, which implies you think stability degraded, hence you saw a problem. But I see this is all theoretical
You are correct. I was saying, "don’t trade stability for speed," especially since I do not user-side perceive that a problem now exists. “Preaching to the choir,” of course, techie-to-techie. I do not perceive any loss of “stability,” now nor earlier.
Now, if you can make this hot-rod go a little bit faster . . . (Isn’t that what hot-rods are for?)
