Spot. On. This is why I haven’t deployed and mucked neck-deep in Discourse yet. I really want to use Discourse, but right now I feel that I am unable to make any subtle changes I need without having a problem down the line if I need to update my server’s instance.
The UI, as is, is great. I know @codinghorror already mentioned that the default theme needs to be perfect since it’s going to be the only theme for a lot of users. I also have seen that the devs are very passionate about some aspects of the ui and that changing them in the core is out of the question.
While yes, the Minimal Discourse edits proved a good point, I truly think only a few elements need to be removed. I specifically speak of the stars and the 5-avatar line in the topic listings. Everything else I think should stay and I think my little mock up from @sam’s initial screenshot with a bit of minor tweaking should replace that 5-avatar bar in each topic listing entry… Maybe others might think differently about these opinions but as it is, there’s plenty of Discourse instances live on the net and I haven’t seen one instance where any other drastic changes were made that would suggest a certain ui feature is probably best to apply to core. The stars is one example of this. No one used the feature and many hid the stars with CSS. That’s been established.
I can also see that technical debt is trying to be avoided by making sure the default UI is solid. But I think this might be a case of paying too much attention to the proverbial bike shed. I experience this in traditional art as “knowing when to stop”. You keep looking at something on your drawing or painting and then go “well, damn, I see this mistake here…” and you keep doing that and picking out new things to change that were trivial in the long run. The original artwork becomes overworked and ruined. One could say the removal of the likes from the topic listings is one example of this.
As @tonninseteli said, this is now a case of personal tastes which can easily be solved by taking this final default ui and making it customizable. In fact, I can count on the latest topic listing a nice sizable sample of discussions that would be completely moot if easier theme/ui customization was available. Even some of my own threads are ui-related and could be solved with easier ui editing capabilities.
Any bugs in the existing ui elements should of course be handled. That I am not contesting.
Any kind of suggestions to cosmetically change the current ui that are not related to serious bugs should probably be deferred until more ui customization options are available.
I do wish to say I am grateful for all the dedication and determination from the devs and contributors and bug testers.
An important few points here that seems to have been lost in “design discussion” is
I made the front page faster (which is part of the goals for our next release)
I demonstrated an involved customisation that people can muck around with today on Discourse instances
I cleaned up said extension (look at edit history)
a. automatic template registration
b. view model for raw templates
c. consistent style
My goal is to make Discourse faster and more extensible both of which are happening. It takes time to nail down extension points and styles but I quite like my oddball design for my own use and can keep probably keep it alive long term for my own use. Which makes Discourse better and more extensible for everyone.
You can muck with my template on your own instance and play with whatever designs you want, which is the point here.
I did not know that. I am glad to hear it was already finished or there’s at least a stable version of it ready for using. Thank you very much and I do apologize if my reply above sounded demanding or nagging.
I think the crux of the issue is that my perfect interface to Discourse is different to what others ideal interface is. We ship a “default” that is a “one size fits all” which @codinghorror makes all the final calls on.
We ship today a site customisation section that lets discourse admins adjust to taste.
We will never be able to get to a global agreement of what the “ideal” minimal design is, but I can, in theory, have a theme that I own and maintain that is my “ideal” minimal design.
Totally agree. The reason I brought it up was that I’ve seen several cases where people asked for or implemented removing something that they thought made something too busy or ugly or whatever, and I’ve disagreed or thought it made things worse.
So I like to make sure I speak up when I see it happening. If nothing else, it’s a minor bump to possibly slow things down by considering that something isn’t a universally held whatever.
I forgot to mention that your minimal layout was a very welcome resource for me as I am looking into adjusting the stylesheets to somewhat similar results. I will most likely do so as soon as I have wrestled this flu, which is partly the reason for my derailing, sorry about that…
As you said, the Custom CSS feature works just like it should and is very powerful. I would simply focus on documenting and commenting the CSS sources or writing a brief reference document instead of developing a wizard to customize site appearance.
I haven’t really been looking into the CSS for more than a few hours, but many of the trivial changes have been frustratingly difficult to track down and implement just by myself.
Edit: Visual finish of a forum might not be what discourse is about, but Discourse is still very appealing to several audiences that value, at least to some extent, appearance over perfect ergonomics, such as designers and online gaming communities, of which latter are the ones most likely in my opinion to benefit from Discourse’s features.
This is easily changeable with CSS. In fact it already happens, see what happens when you view a topic list with a portrait iPad or similar. (Or make your browser very narrow by dragging the edges with a mouse.) That is all CSS as well.
Stars are definitely coming out, in fact now that we finally made the call, I wonder what is taking @sam so long to remove them
Glad to hear this! Though I admit I am not the type to do a display:none and wash my hands of it. When I do that, I am merely hiding it but the feature/function still exists in the core itself. I prefer things under the hood to reflect. This is a pedantic notion and I understand that. My finickiness will be the end of me one of these days.
My main concern was being able to easily find where one left off reading. But as that isn’t going to be a problem (because I know I’ll need to click the topic title) I’m fine with the decision.
One of the obvious truths I was reminded of - years ago as I watched a child go through a school system - is that different people learn stuff differently. It was a lesson useful years later to creating Powerpoint slides: some like pure text, some appreciate diagrams, and others like Steve-Jobs-esque evocative backgrounds with few words. My bias? I like this particular minimalist look because it helps me get work done more quickly: I get the names for topics, whether there’s been any response, and the poster’s (replier’s) names. For me, it’s the important information I need to process what’s happening and judge what to do next – without a string of visually distracting avatars.
This is disingenuous, it completely burries the functionality. I also wonder how this would alter the topic view, as right now you can click the star in the title and get the current topic starred.
Compare this to GMail: when they added the “Important” marker, they added it back, front and center; each time it can be clicked to toggle the state. They even added a “Mark as (not) important” in the “e-mail thread actions” menu so that it would be more discoverable.
With this new design, it seems you would need to click the topic (and read some posts if there are any new unread ones because AFAICT you can’t go straight to #1 from the topic list), use the post counter to go to “Top” and bookmark that first post. This does make it feel like jumping through hoops…
Great - works like a charm. This helps greatly to fiddle around with the default template - without having to dig into the built-in handlebar templates! I really feel like this should ship with Discourse by default…
Is there any overview over the hooks you can use or overwrite to create your own template like this?
No. These are useful, with maybe the exception of Top. They represent different things which are useful to have split out.
I’d rather this didn’t happen as being able to track if something is both starred and unread or new is useful. Maybe have a poll, see who agrees across the customer base?
Please don’t. Even over there where avatars change more frequently, it’s a much easier way of quickly seeing who the top contributors (and OP/LP) are without worrying about it. This is useful to me. TLDR on this and the next point - images/colours are easier to distinguish at a glance than words.
The colours of the categories helps split them out without paying any real attention to the text of the label after a while. Again, I find this useful.
1 post vs 0 replies just seems like semantics to me. Maybe this makes a difference to some, worth a poll?
There are actually somethings that Discourse gets right compared to other forum software, and some of them are listed here as candidates for removal. Please, please, please either leave them be and let people who don't want them remove them with CSS, or make them admin options on a per-site basis. Don't just make them disappear.
Keep in mind, this is my design we are talking about for my personal use, not a global change, as the title says … a hypothetical simpler minimal design.
The only thing on the table at the moment is removal of stars from the topic lists which reduces visual clutter and unification of stars / bookmarks
Though I admit I am not the type to do a display:none and wash my hands of it. When I do that, I am merely hiding it but the feature/function still exists in the core itself. I prefer things under the hood to reflect. This is a pedantic notion and I understand that. My finickiness will be the end of me one of these days.