wiki homepage (was: noob questions about the wiki)

User avatar
Daniel Beardsmore

08 Jul 2014, 23:12

It's far more complicated than that. Remember that rabbit hole? Let's go inside.

(As an aside, I have no idea why you can't browse to companies in my proposal — I only added four boxes to the page as deep down I knew I'd be shot down, so I didn't sit up all night drawing icons that would never see the light of day; the original sketch has more of the boxes in there, and for goodness sake you can have as many boxes as you want: that's what MediaWiki templates are for!¹)

Let's start with: [wiki]Alps SKCM White[/wiki]

See the keyboards section? That's hand-generated. Let's assume we had a category path as follows:

Category:Keyboards with Alps switches → Category:Keyboards with Alps SKCL/SKCM switches → Category:Keyboards with Alps SKCM White switches

Adding a keyboard to that deepest category does not add the keyboard onto the switch page automatically. Doing that is physically impossible as it stands, as category membership doesn't include qualifiers, e.g. some of keyboard X has switch Y. Category browser transclusion wouldn't in of itself be too hard to achieve. Note that that specific page has some custom HTML/CSS to get the columns otherwise the list would get too long for the page; I had to compromise the layout as CSS columns suck (WTG W3C).

Category membership would not permit you to list a keyboard solely by its name plus an external reference; it would force a page to be created just to hold that reference, so you'd have to click through to get nothing more than a hyperlink. In an ideal world it would encourage someone to populate the page, but they won't and you know it. I've lost interest in the wiki for the most part — I still update it, but I've flogged the dead horse so much I've broken the whip and ground my bones to dust.

Another one: apparently only three keyboards in the world use MX brown: [wiki]Cherry MX Brown[/wiki].

The level of decentralisation in a wiki used as a catalogue is pretty bad and all the pages are constantly out of sync with each other.

These pages are forgotten and don't ever stay current:

[wiki]Keyboards and switches by year[/wiki]
[wiki]Keyswitch timeline[/wiki]

The only way to get out of this is a fresh object framework system that embeds a copy of the MediaWiki engine solely for the free text content.

As it stands, your argument seems to be in favour of poky boxes on the front page where nothing fits, that are far too arbitrary and incomplete, have no explanations, are very confusing to newcomers, over having a friendly link people can click through to a page that does the exact same thing as the box, but with room for explanatory text, icons/images for each sub-section, and clearer lists.



¹ I might even have a template to convert red/green/blue/grey to the CSS colours to save hard-coding those and requiring mental gymnastics to recognise and swap the colour codes when moving boxes, although nth-child CSS could also take care of that directly

User avatar
webwit
Wild Duck

09 Jul 2014, 01:25

If you want to get technical (this is part of what I do for a living), when someone visits a website, in most cases the visitor has a direct information goal. Example: search info for Cherry MX. I think what's good about your first draft proposal is that it would fulfill a great percentage of the visitors' information goals with a minimal of noise. So you have something going there. But these are not all information goals or modes of fulfilling these goals. and this is what has been wrong about this discussion. It's partisan in the way that this person likes it this way and that person likes it that way, while the trick is to make it work for everyone in both direct search/hit or in discovery mode.

The solution you propose, content pages which manually point to deeper content, are still more high maintenance than the category system you say doesn't work because people won't bother. I don't think we disagree that much in that we both know there won't be many people actively maintaining any structure. But I think you dream about a perfect solution while I search for a realistic one. You point to problems of the homepage due to organic growth and lack of refactoring but propose something which will have many such pages in the end. It replaces one system which is imperfect with another which is imperfect for the same reasons as in requiring strong editing, which will end up in two imperfect systems, adding to the confusion. This is a small hobby. Even less are interested in documenting it. What is realistic with these realities in mind? I don't claim to have the definite solution.

User avatar
Daniel Beardsmore

09 Jul 2014, 09:37

Quite honestly, I don't care about this strongly enough, so you win this one. (I don't agree with you, but I can't be bothered to argue.)

User avatar
matt3o
-[°_°]-

09 Jul 2014, 09:39

sorry guys I was swamped yesterday. I'll work on some suggestions later today... after I read the various "wall to text"...

User avatar
Madhias
BS TORPE

09 Jul 2014, 09:55

kbdfr wrote: In my opinion the existing structure is about the best to be achieved. Clear, simple, full of information.
I also think so. It's a Wiki, so I don't see any problem with it. Of course I can't discuss here that much, since I do to have contributed to the Wiki yet.

Most important is to find information fast and easy. And that is the situation now.

User avatar
matt3o
-[°_°]-

09 Jul 2014, 15:57

Of course the images are place holders, of course we still miss some material, of course is super draft, but this was the idea

Image

Little less info, better organized.

Take it, leave it. I don't care.

User avatar
Muirium
µ

09 Jul 2014, 16:00

Nice! Reminds me: I'd have a good description of the meaning of linear, clicky, and tactile given high profile in the switch section. They are not as obvious terms as we think. Also, Keyboards section deserves first position in the top left. They are the reason for switches and caps to exist.

User avatar
matt3o
-[°_°]-

09 Jul 2014, 16:05

looking at wiki stats, switches are far more important than keyboards (reason why I put them first)

User avatar
matt3o
-[°_°]-

11 Jul 2014, 08:24

it seems all my great expansion ideas die in the oblivion :(

I have another project coming... I haven't spoken about it here because you are such low-ballers :P

User avatar
Daniel Beardsmore

11 Jul 2014, 09:55

I should put the Nike logo on the front page of the wiki: my rule is "Just Do It". Once I'm satisfied with a plan, I'll just do it. I don't waste time consulting anyone else at Deskthority — I never get any answers and, largely, no-one seems interested enough to care whether I changed anything or not when I actually do do it. I do of course realise that not everyone can be Steve Jobs: you can be extremely passionate about a completely wrong idea, as so many people are. However, this is a wiki: you can roll everything back if it comes to the worst and everyone hates what you did. Odds are 99% that nobody notices or cares; you take home only the personal satisfaction of having made something better in a way that no-one has the ability to recognise, but will nonetheless be better off in the long run.

Another reason I'm wary of front page boxes full of text is that CSS sucks. Your mock-up shows boxes that are all different heights. Now I realise that a CSS property such as height: greatest would be painful. For example, the rule "div { height: greatest }" on a page with nested divs would crash the browser, so you'd need to ensure that height calculations are cascaded up from innermost to outermost element. Another pet peeve is that you can't have character-separated elements (e.g. "Keyboards | Mice | Switches …") and not have the separator characters appear by mistake at the start or end of a line, when the text wraps.

I've been concerned about the front page for a while. However, it's not bothered me enough to decide to rebuild it. I don't know how most people reach the wiki, but I imagine more of them arrive through web searches and from forum links than by clicking the link within the forums to reach the front page.

My feeling personally is that a reference site is only useful if the odds are high that what you're looking for, is covered. It's one reason why I always start out with Wikipedia — everything you can imagine and more is covered. Not quite, but by and large, I don't need to ask myself whether I'm likely to find what I want there.

The wiki coverage is very low in many areas that really count: Ducky's product range for example is largely missing, as is the new Kul brand, much of Filco (just product lists, no pictures or details), Leopold, WASD (the WASD Keyboards page still reads "The following are in the company's official roadmap for 2012 …"), KBT, KBC/iKBC … (Pure, Pure Pro, Poker, Race, etc) The Topre page is just a massive table of models with no pictures of them that will scare most people to death. HHKB coverage is still lacking, though someone is/was actually working on that.

It doesn't matter how nice the front page is, if almost every major keyboard brand is not covered — the fantabulous new front page won't lead anywhere but dead ends. You say switches are more popular? That's possibly because they're covered in far more depth and detail than current on-sale keyboard products. You won't be getting hits for keyboards that aren't on the wiki — there's no page to hit.

Good coverage means that people can build an instinct to use the wiki as a first port of call for knowledge. It means that people simply searching Google/Bing/etc will encounter our wiki. Good illustration means that people who can't read English will go there even if they realise that they can't get a readable translation: that's why Sandy's website is so valuable: pictures, pictures, pictures, pictures, pictures! There's a picture of pretty much everything, so even if I can't read the page, I can see what I need to see from the pictures, and I can then tie that in with a terrible machine translation.

User avatar
002
Topre Enthusiast

11 Jul 2014, 11:14

Daniel Beardsmore wrote: The Topre page is just a massive table of models with no pictures of them that will scare most people to death.
You spew some shit sometimes, Daniel. I am assuming you're talking about the Realforce page which has quite a decent write-up that took me some time and research and you just write it off like it's a useless page that no-one could possibly like. What value do you see in 200 pictures of Realforce models where the only difference is a couple of keycaps?

User avatar
webwit
Wild Duck

11 Jul 2014, 13:04

I never get any answers and, largely, no-one seems interested enough to care whether I changed anything or not when I actually do do it.
I tried to engage in an intelligent discussion and your answer was you couldn't be bothered.

User avatar
matt3o
-[°_°]-

11 Jul 2014, 20:14

being my day work I can comment on this:
Daniel Beardsmore wrote: Another reason I'm wary of front page boxes full of text is that CSS sucks. Your mock-up shows boxes that are all different heights. Now I realise that a CSS property such as height: greatest would be painful. For example, the rule "div { height: greatest }" on a page with nested divs would crash the browser, so you'd need to ensure that height calculations are cascaded up from innermost to outermost element. Another pet peeve is that you can't have character-separated elements (e.g. "Keyboards | Mice | Switches …") and not have the separator characters appear by mistake at the start or end of a line, when the text wraps.
CSS sucks, we agree, but both same height elements and good looking separator are possible. I have just to see how this wiki works.

I will follow your suggestion and simply push my suggestions to the wiki and see the reaction.

User avatar
Daniel Beardsmore

11 Jul 2014, 22:29

002 wrote:
Daniel Beardsmore wrote: The Topre page is just a massive table of models with no pictures of them that will scare most people to death.
You spew some shit sometimes, Daniel. I am assuming you're talking about the Realforce page which has quite a decent write-up that took me some time and research and you just write it off like it's a useless page that no-one could possibly like. What value do you see in 200 pictures of Realforce models where the only difference is a couple of keycaps?
Does that in any remote way diminish my argument? Where are all the pages on keyboards from Kul, Leopold, Ducky, KBT, iKBC, WASD, etc? Someone the other day asked if he could swap Fn with another key on (IIRC) a Ducky Shine 1. I can't look up the Shine 1 and see whether it had any DIP switch or programmable functionality — the only pages so far are the horrible 1000-series models, which if you believe the wiki, are still on sale.

It's meaningless to argue that the switch pages are more popular than keyboard pages when the keyboard pages that should exist, simply don't.

User avatar
Daniel Beardsmore

11 Jul 2014, 22:44

webwit wrote:
I never get any answers and, largely, no-one seems interested enough to care whether I changed anything or not when I actually do do it.
I tried to engage in an intelligent discussion and your answer was you couldn't be bothered.
It was going around in circles. Once I've stated my position, there is no value in repeating it over and over; I have nothing further left to add. cf Monty Python argument sketch. I can't change your opinion. I can't change anyone's opinion. I never could. If you imagine a scale that terminates in the Steve Jobs level of person who can sway anyone, I'm diametrically opposite to that.

As I said before, the wiki is a "Just Do It" system, and if I still cared enough, I'd just wipe the front page and start over. Someone would naturally kick up a fuss, like 7bit when I ruined his beloved, precious SGI Granite page; the irony is that, despite all his protesting, he never kept his side of the bargain and never created the SGI Granite page he wanted, so I just cleared it and pointed it back at the correct page, since he left it as a confusing duplicate of another page.

Yet if you do ask for advice, you'll never get any agreement and consensus, or even much interest; all you're able to do is follow your heart and do what you were going to do anyway, since most people couldn't care less.

I don't intend to change the front page; the wiki is a dead horse that I've already flogged way too much, and the front page is so far down the list of real problems, it's not something I ever gave much thought to. I do agree that it needs redesigning, but I feel that the priority is having a much more complete subject coverage, including the quality illustration that people here are capable of (the forum is filled with wiki-grade images). Something that will bring in more visitors and radically improve its usefulness as a reference source.

As such, I conceded that debating the front page design was completely futile.

User avatar
Daniel Beardsmore

11 Jul 2014, 22:54

matt3o wrote: CSS sucks, we agree, but both same height elements and good looking separator are possible. I have just to see how this wiki works.
In case you're uncertain, the main page uses a specific template for the blocks that you can see here:

[wiki]Template:Main_block[/wiki]

There's no example given, but the front page demonstrates how to use this. For reasons explained in a second, you should probably create a new template for the new blocks.

MediaWiki allows full HTML and CSS, although it's good practice to use MediaWiki markup whenever possible, for ease of editing.

If you need to amend the global CSS, edit this page:

http://deskthority.net/wiki/MediaWiki:Common.css

I don't know what most of it does, but there's a section at the end labelled /* Deskthority custom css */ which is what you want. The main page block template already relies heavily on CSS code defined in Common.css, as do some of my other templates.

MediaWiki has a limitation that accessing a previous version of a page does not access the contemporary states of images, templates, CSS etc, so I presume that you need to preserve all the existing CSS for the existing main page, so that people reviewing its history will still have the required CSS present. Personally I think that browsing a page's revisions should use the contemporary versions of all the templates and images; there's a lot wrong with MediaWiki.

Hence the need for a new template, otherwise viewing the older revisions will completely break. (Not unless one of the wiki updates has finally fixed this gaping hole.) I found this problem out when consolidating the infobox templates for keyboards and switches.

If you have any other questions, just ask, or send me a PM. I'm always happy to help people learn MediaWiki.

User avatar
matt3o
-[°_°]-

11 Jul 2014, 23:53

thanks for the insight. I'm renewing my hate for this system :) but I'll give it a shot, maybe at the end I'll like it.

I will have a look at it this weekend, I'll let you know if I need help. you will surely have to review the modifications I make.

Post Reply

Return to “Deskthority wiki talk”