glitch

I can't actually tell if that's sarcasm anymore or not. Next slide is a new URL, new page. Based on initial testing, the behavior seems fine. Same behavior as one would have for something like http://example.com/my-presentation/1.htm, 2.htm, 3.htm, etc. and navigating them with hyperlinks on each slide that link to neighboring slides.

It's not like Back/Forward browser buttons are overrode to behave and previous/next for the slide presentation.

Remember those horrible embedded Flash presentations that you couldn't directly link to a particular slide within the blob? Yeah, that "breaks the Web". Back/Forward is supposed to go back to the previous page the user was on (which is a "slide" in this case).

Using Chrome 43.0.2357.81 (64-bit) / OS X.

Just to add to this discussion, some quasi-related links with regard to visual presentation of code:

• Experiments in Code Typography (McDirmid, Microsoft) http://research.microsoft.com/en-us/projects/liveprogramming...

• Typography of Code (MAX 2010): https://youtu.be/mG0lyGekGDs (alternate recording: https://youtu.be/r2JePjrDggE)

• Elastic tabstops: http://nickgravgaard.com/elastic-tabstops/

Of course, there's also numerous projects with more significant visual representation, like Blockly, Scratch, flow programming, etc. And, of course, real-time feedback/simulation is another arena of usefulness e.g, http://research.microsoft.com/en-us/people/smcdirm/managedti..., Bret Victor's works, etc. But, I digress.

"4502" might be a mushing of "6502" and "4510" together, making a name for their own thing? (The 4510 is a system in package variant of the 65CE02.) I too am not familiar with any "4502" in this context.

Admittedly the first sentence in the README.md points to relevant information (http://setosa.io/ev/) However, yes, they should have included the one sentence description on that page it links to in the README.md itself.

Explained Visually (EV) is an experiment in making hard ideas intuitive inspired the work of Bret Victor's Explorable Explanations. See http://worrydream.com/ExplorableExplanations/ to learn more about Explorable Explanations.

See "Media for Thinking the Unthinkable" (Bret Victor, 2013) https://vimeo.com/67076984 and videos from https://vimeo.com/worrydream/videos for more context. Also, http://explorableexplanations.com (by Nicky Case) is a.... page that exists.

They could do something along those lines, but that would not be "keeping things simple". Plus, there is not significant motivation for them to do it — plain and simple. They could, but so far it's not worth the time/money investment. Every little feature, every little bit of work added and layered onto things has an associated cost/benefit to it. Simply put, it's not worth it to them at this present time.

(1) Keeping it simple and consistent and (2) they want to ensure that as much relevant traffic goes through t.co as possible. If you want to lock people into t.co URLs, there's no reason for them to play games otherwise unless it truly starts eating into the character count. It's an avenue for profit through data mining. Also, the notion of blocking phishing sites is also there, but the real reason is data mining. They want as much desired, relevant data as possible about the URLs people are using on their service — where they're going, when they're going there, who's going there, etc., etc., etc. In the data, there is money. It's also about other untaped potential in being the middle man (e.g., injected ads) — not that they'll necessarily do such a thing, but from their perspective it's good to be in the position to exploit as many avenues of profit as possible.

> Partial data will serve as a distraction if you present it as important.

If you ascribe to the importance of its display, then you are allowing it to be a distraction as an observer of the display. But, yes, in practice I agree that most humans are foolish and will allow it to serve as a distraction.

> I'm saying this doesn't belong on a dashboard. Granted, they just did this for fun.

Maybe or maybe not. There are definitely good reasons for removing it from the dashboard. I agree that it is not presented well. Outright permanent removal of the vague notion and concept of this particular data and its relevancy is a different matter. Exploring alternative presentations, getting more sensors and more relevant data, etc. (This is of course stating the obvious.) But I'm not really commenting on that at this time.

> The more data the better - That's not true. Some data is better than others. Some data is poor quality. Some is unimportant.

This is often debated with vigor on both sides (okay, there's actually a gradient of n-dimensional in-betweens, not explicit sides) of the argument. (Collect it all and figure out relevancy from analysis, then scale back data collection where it was determined to be irrelevant. Be presumptuous based on experience and collect only what you believe is relevant. Expand and contract collection periodically as something in-between. The list goes on and on and on and on of possible configurations and approaches.) Of course some data is better than others. Some data is more relevant than other data for a given operational context. And, yes of course, besides just relevancy, fundamental things like accuracy, precision, and error are all aspects of data collection and analysis. But, back on topic, both sides of the argument have merit. The reality is, like all things in life, there is a pragmatic balance point. (Not necessarily a point, though. In reality, usually an n-dimensional volume, but that's another topic.)

Such is the innate aspect of data gathering and analysis. The more (relevant) data, the better. (What determines relevancy is not discussed here for brevity.)

Partial data will only serve as a distraction if you allow it to. EDIT: If you falsely present partial data as important, yes, then it very well probably is distraction because by you are facilitating it to be a distraction by virtue of presenting thusly. As consciousness, you should try to be aware of inappropriately disproportionately presented data expressed by others in your environment and to not let it serve as a distraction to you as an observer. For example, Fox News and how it goes about its interactions displaying information (true information, false information, incomplete information, etc.) with the masses is a popular subject for case study in this type of scenario.

If you are aware that it is an incomplete picture, then you should be fine. Again, the more data, the better. Be aware of limitations and incomplete pictures. Be aware of context. Same old story of much everything in life. Always be exploring further. Always keep on questioning what's happening in the environment and the universe. Ignorance is bad. Of course, balance is needed. Pros are something that exist. Cons are something that exist. Things exist. Actions exist. Interactions between things and other things with actions exist. Systems of interaction exist. It's turtles all the way down. But yes, based on history, most humans fail at comprehending their own basic existence, their interactions with others, and anything else contained within their neck of the universe. Rather depressing actually. Nothing new here.

EDIT: As far as what belongs or doesn't belong on a dashboard, that would take a long time for me to discuss. In this particular example provided in the original posting, yes, it has various flaws. Not discussing that.

Also, with regard to "the more data, the better", yes, irrelevant data has negative consequences and must be dealt with accordingly. Of course, the a perspective exists where you should try to identify irrelevant data and not gather it to begin with. Conversely, there is a perspective where you should gather it all so as not to presumptuously exclude and that the data itself will tell you what is relevant and irrelevant during analysis. Both approaches express certain positive merits and also negative qualities. Identifying what is irrelevant and what is relevant has its own intricacies which are not discussed here in any relatively significant capacity.

> But how do you know which interactions are useful to your business and which are not? How do you the right amount of flow through an area? What about the things you can't measure, but may matter to the health of your workers?

For most people, using their experiences throughout their life as a perspective. Once you get into a new workplace, you can use those years of experience as a basis for perspective relative to the observations of the new workplace. For a lot of people, their intuition makes this kind of thing obvious once they start looking at the data and look around casually in the workspace with note of looking at the people, their movements, and what they're doing. It's the kind of thing that after working at a new place for a week, I already would have a solid feeling and a useful perspective — not just from the new workplace, but comparing that new workplace based on my experience of other workplaces that I've worked in, consulted for, and otherwise visited.

But, there are formal studies on human interactions in the business workplace addressing all three of these questions. I don't have particular citations of interest off-hand.

> I still don't see how knowing stair steps in isolation will generate any valuable insight.

Strictly empirically speaking, it doesn't really. However, this set data from one location combined with observational perspective of the people within the space, could easily lead to more informed conjecture and related hypotheses.

However, the novelty of the single location data set may be as useful distraction; to serve as an element introducing "fun" and "amusement" into the workplace at acceptable levels of "fun" and "amusement". (There are studies on this, as well. This information can serve as a device to break monotony and a "humdrum" atmosphere. You can employ other devices as well. For example, free soda can be used as a device. Whether or not that device works effective and the full consequences (positive & negative) is beyond the scope of this simple comment.)

If you want to pursue this beyond typical intuition, I recommend taking human behavior studies in psychology with respect to interior design and architecture.

This goes beyond just traffic flow, but incorporates other aspects of the environment and its influence on humans.

However, you don't need to have academic training to be useful. Many people have the intuition to reasonably and appropriately make observations, analyze, and comment on their surroundings.

But, on the more academic side of things, here are some quick and dirty Google search results. Random examples of literature in ballpark of this field in an assortment of contexts:

A Homely Approach is Paying Dividends for Patients and Staff. (2012). Nursing Standard, 27(2), 20-21.

Rydeen, J., Erickson, P., & Lange, J (2008). Built for Brains. Industrial Engineer: IE, 40(3), 32.

Joye, Y. (2007). Architectural Lessons from Environmental Psychology: The Case of Biophilic Architecture. Review of General Psychology, 11(4), 305-328. doi:10.1037/1089-2680.11.4.305

Read, M., & Upington, D. (2009). Young Children’s Color Preferences in the Interior Environment. Early Childhood Education Journal, 36(6), 491-496. doi:10.1007/s10643-009-0311-6

Moore, G. Architecture and Human Behavior: The Place of Environment-Behavior Studies in Architecture. http://sydney.edu.au/architecture/documents/staff/garymoore/...

Cummings, N. Fostering Sustainable Behavior Through Design: A Study Of The Social, Psychological, And Physical Influences Of The Built Environment. http://scholarworks.umass.edu/cgi/viewcontent.cgi?article=20...

Esmaili, Y. Consideration Of Human Behavior In Designing A Green Office Space For Environmentalists A Simulation Study. http://architecture.arizona.edu/sites/default/files/faculty_...

The stairs maybe singularly interesting as a novelty or curiosity, but to gain better perspective, you'll want the stairs, hallways, entry/exit-ways, etc.

If you can gather data of human traffic flow around the workplace in all sort of spots, you can then analyze that data to gain perspective on that flow. And it's not so much of the obvious patterns, but maybe those that didn't seem obvious (until hindsight after the fact). Proximity and placement of different facilities (printing/copying facilities, toilet facilities, lounge/kitchen facilities) within the office space can lend to different interactions or even wasted time.

And, of course, one can casually observe and make qualitative observations. But those qualitative observations may not necessarily be impartial nor rigorous in their recording. To avoid subjectivity and anecdotal evidence, quantitative evidence through raw data recording aids in eliminating potential subjectivity.

Based on this, you may choose to layout the office space differently. Or when choosing a new office, you may better know certain attributes you're looking for in a space.

For example, Pixar's centrally located toilet facilities intentionally shape traffic flow with the intent of increasing interactions among employees for facilitating potential conversation and discussion.

As shiftpgdn mentioned, another example seen often is stairs vs. escalator/elevator usage for encouraging subtle, healthier choices. The counter, when displayed publicly for those respective areas, can serve as a mechanism for real-time, interactive feedback to the observer. Confronted with this reminder, it may coax them a little, giving them the little mental kick to use the stairs instead.

Maybe you determine that if you put an appropriately shaped obstruction — e.g., a wall with a sign of company logo — carefully placed immediately in front of the entryway, it may result in better flow through the that area. Notably, it might encourage people to not stop in the doorway, blocking; but rather proceeding further into the lobby area to converse.

Although more of a larger crowd thing in this example, studies have been done in this area and have demonstrated such results: http://www.americanscientist.org/science/pub/strategically-p...

Related reading:

"Handbook of Positive Behavior Support" by Sailor, et al.; "Behavior Supports in Nonclassroom Settings" chapter, "Features of the Physical Environment" section (page 499).

"Office Planning and Design Desk Reference" by Rappoport, et al.

Agreed. The article writer really failed in that section. It's just rounding error. Based on his written reactions there, I honestly don't think he realized it was oscillating at the time he wrote that part. It seems like he just didn't want to bother to correct himself after the fact or is just intentionally being stupid.

As King of Sun III, yes I am forcefully "volunteering" my subjects to do the work. Archives can remain as-is. Everything new shall use the new system. Doing otherwise incurs penalty. If you don't use the new system, you will get pain treatments and possible death. /humour

Yes and no.

Yes, additional sizes address the desire for satisfying the missing in-between sizes.

No, because I'd like to keep it simple and avoid having to do that or even think about it at all.

Basically, with the ANSI sizes you want to satisfy TWO proportions with gradual size changes instead of just ONE proportion with gradual size changes as seen in ISO 216. This is where the discussion cycles back to the utility of 1:√2.

In most offices in the U.S. that I've been in (I say this in a fast and loose fashion), they carry three sizes (with the following typical uses):

(1) ANSI A (US Letter) for standard letters (uses full portrait page) and leaflets (uses one-third landscape page) or booklets (uses half landscape page).

(2) US Legal because there are plenty of legal documents are out there that use US Legal (uses full portrait page). And perhaps as a bonus, but not as frequently seen, US Legal can be folded into small booklets (uses half landscape page). Although, I don't see this happen as much as ANSI A being used for booklets.

(3) ANSI B (Ledger/Tabloid) because it can be great for spreadsheets and folded to US Letter sized pages. Either for collapsing the spreadsheet or for making booklets with US Letter sized pages (uses half landscape page).

Now, yes, I did not mention all uses. I merely mentioned what I have frequently seen used based on my experience. But one of the key take-away points for the "common-use/general-purpose" ANSI A and ANSI B paper sheets is the notion of folding the paper in half to form a booklet.

You don't see many offices carrying ANSI C (17"x22"), which is the next step up from ANSI A maintaining the same proportion. (So availability of this is considered unlikely in this context.)

Of the three, perhaps US Legal can stand alone because it offers a significantly relevant special situation proportion. So, we'll set US Legal aside for now.

And so, you get usage of Half US Letter (1.5454…), Full US Letter (1.2941…), Half Tabloid (which is just US Letter, 1.2941…), Full Tabloid (1.5454…). So, you'll get an oscillation of two distinct proportions with paper that doesn't use 1:√2, and in the case of ANSI A through ANSI E, you get the 1.5454… and 1.2941… oscillation.

Now, if you had ISO 216 sized paper, all of the proportions would be the same no matter what, and it becomes very easy to choose between paper and half-paper sizes because they all work alike with regard to their proportions.

However, with the oscillating ratios, you have to choose which one you're targeting and be sure of it, because the next larger (or next smaller) size is significantly different.

Now, more in-between sizes could solve this, but that also means you'd probably end up carrying around yet another paper size because of it. If you simplify things by getting rid of the dual-proportions and have a unified proportion instead, then the added frustration from the differing layout proportions go away instead of trying to address that with another paper size for something in-between.

So, to parallel the Half US Letter (feels like an oddball that is basically super small Tabloid, 1.5454…), Full US Letter (1.2941…), Half Tabloid (is just Letter, 1.2941…), Full Tabloid (1.5454…):

Half A4 (is just A5, 1.4142…), Full A4 (1.4142…), Half A3 (is just A4, 1.4142…), and Full A3 (1.4142…).

With the layout for Half US Letter, your next proportional step-up is Full Tabloid. (Huge jump!) Or from Tabloid, your next proportional step-down is Half US Letter (Really small!) You could have an in-between size to satisfy it, but it seems excessive when you could have just had used the 1:√2 proportion to begin with and not worry about any of those additional considerations.

---------------------------------------------------------

Saying it another way...

Where, ½·x signifies folding a sheet for purpose as a booklet, and the paper that satisfies that requirement is mentioned inside the cell.

Dealing with only our "letter" size (1) and "tabloid" size (2) sheets that we have in our office, we get the following:

    ANSI    | ½·1 | 1·1 | ½·2 | 1·2 |
            +—————+—————+—————+—————+
    1.2941… | :(  |  A  |  B  | :(  |
            |—————|—————|—————|—————|
    1.5454… |  A  | :(  | :(  |  B  |
            |_____|_____|_____|_____|


    ISO 216 | ½·1 | 1·1 | ½·2 | 1·2 |
            +—————+—————+—————+—————+
    1.4142… | A4  | A4  | A3  | A3  |
            |_____|_____|_____|_____|

With ANSI:

    We don’t have a “half size” booklet with 1.2941… proportions.
    We don’t have a “full size” sheet with 1.5454… proportions.
    We don’t have a “full size” booklet with 1.5454… proportions.
    We don’t have a “double size” sheet with 1.2941… proportions.

With ISO 216:

    We have our “half size” booklet in all our proportions.
    We have our “full size” sheet in all our proportions.
    We have our “full size” booklet in all our proportions. 
    We have our “double size” sheet in all our proportions. 
    (Because we only have one proportion to deal with, 
    there’s no bothering to deal with dual proportions.)

That's four frowning faces (missing solutions for that proportion) on the ANSI side.

---------------------------------------------------------

If we were to expand on this beyond the "two typical sizes found in the office" scenario, we get the following:

    ANSI    | ½·1 | 1·1 | ½·2 | 1·2 | ½·4 | 1·4 | ½·8 | 1·8 | ½·16 | 1·16 |
            +—————+—————+—————+—————+—————+—————+—————+—————+——————+——————|
    1.2941… | :(  |  A  |  B  | :(  | :(  |  C  |  D  | :(  |  :(  |  E   |
            |—————|—————|—————|—————|—————|—————|—————|—————+——————+——————|
    1.5454… |  A  | :(  | :(  |  B  |  C  | :(  | :(  |  D  |  E   |  :(  |
            |_____|_____|_____|_____|_____|_____|_____|_____|______|______|


    ISO 216 | ½·1 | 1·1 | ½·2 | 1·2 | ½·4 | 1·4 | ½·8 | 1·8 | ½·16 | 1·16 |
            +—————+—————+—————+—————+—————+—————+—————+—————+——————+——————|
    1.4142… | A4  | A4  | A3  | A3  | A2  | A2  | A1  | A1  | A0   | A0   |
            |_____|_____|_____|_____|_____|_____|_____|_____|______|______|

That's 10 frowning faces on the ANSI side, and no frowns on the ISO side. The ISO side inherently avoids the frowning face scenario.

Yes, you could add on another set of sizes to address the missing proportions on the ANSI side. But, that's inelegant and unnecessary if you just went with the ISO way instead.

It is a big, diverse world. And various, special use-cases are best addressed with different aspect ratios. But, to me, the point of ISO 216 adoption in place of US Letter (and other common sizes) is about having a general-purpose/common-use aspect ratio with inherent properties that help simplify things across the board for a lot of people. If it wasn't already used so much elsewhere, I'd be a "harder sell" (even to myself). But, it is used elsewhere... a lot... and has beneficial properties too. Trying to clean up the mess and simplify the chaos. Remove the unnecessary extra baggage associated with other systems.

Aside: If I was king of Sun III (previously known as Earth), my replacement for "US Legal" sized paper would be 210mm × 340mm. This maintains the width of A4, but gives it a length similar to that of US Legal. Moreover, it achieves this using the Golden Ratio. Calling it φA4 for the time being.

    US Legal..:  8.5”  × 14” (Before mm conversion.)
    US Legal..:  216mm × 356mm —> approx. 1.647:1 ratio

    A4........:  210mm × 297mm —> approx. 1.414:1 ratio
    φA4.......:  210mm × 340mm —> approx. 1.619:1 ratio

See http://i.imgur.com/S0rrRnP.png for an example.

Aside: Apparently the 210mm × 340mm size is used for some folded paper towels and some pastry bags already. Also, Mexico's "Government-Legal" is 216 mm × 340 mm — same height as my proposal but with a 6mm width difference. (Theirs matches US Letter/Legal width of 216mm instead of the A4 width of 210mm.)

> 3) [...] the ANSI paper sizes simply have two aspect ratios alternating as one increases paper size.

In relatively big jumps (of the same proportion), though. Not enough in-between sizes (of the same proportion).

Additional note: Yes, I realize 17"x22" (ANSI C) is a thing. But, that paper wasn't readily available in the specific scenario given. The point is that if all common-use paper has the same proportion, then availability of needing "this or that certain proportion" is less of an issue... or rather becomes a non-issue. When all commonly used paper has the same proportion, then the possibility for the inconsistent proportion problem occurring in every-day life essentially goes away.

Makes it easy for the "large print" edition for people with eyesight issues.

To answer your question about my preference... For non-fiction books, A4 pages seem too big for me, A5 pages seem little too small to me, and B5 pages seem just right. Again, this is just for non-fiction books. Fiction books I prefer more A5-sized. Just my own personal, subjective preference. That's all.

Lack of demand in the States for ISO 216 sizes in general makes it so that relatively/comparatively few stock such paper in the U.S. When wanting to print to the edges (full bleed), publishers will often print on B sized paper sheets with crop marks for A size (and then crop the paper to A size) or similarly print on A and crop to B.

Although I don't agree with the width statement, I just wanted to note that I personally find the B sizes nicer than the A sizes for non-fiction books. I find A5 pages a tad small for textbooks, but B5 seems just right. (So, effectively "B4 folded", which is usually printed on A3 to begin with, then folded into signatures, bound, and then ultimately cropped.) And since it's all the same proportion, it makes it easy.

I really wish we'd switch to the ISO-216 sizes. A4 to A3 is so much more enjoyable and painless than Letter to Tabloid. This kind of thing came up often while making posters for campus because it was cheaper to NOT print out all of the posters on Tabloid. Rather, using a mix of Tabloid (for the important poster posting spots) and Letter (for unimportant spots and as flyers for handout). Always ended up having to spend more time to make two layouts because of it. And A4 to A5 for booklets without having to re-layout the document is a huge plus too! And for books, I like to use B4* to fold in half. (B5 pages are nicer than A5 pages for the kind of books I tend to read.) The proportions are all the same, so it just works! One layout, many paper size (of the same proportion) possibilities!

Edit: Additional note with regard to the book publication (depending on printing and binding method used)... Sometimes/often this is printed on something like A3 size for bleeding and with crop marks. The A3 paper is then folded and cut, so that the pages become B5 size. (Signatures essentially composed of folded B4 as a consequence of the cropping. There is of course difference between the outermost sheet and the innermost sheet in each paper signature, which is why it's all cropped at the end.)

Yes, I agree with you in essence. I don't agree that it's pointless, though; I believe there exists some merit to it. The point of the comparison was more about a quick and dirty quick example with the compression implementation and settings used by GyfCat, which was a quick way to convert the GIF. There are obvious flaws in this approach. Many flaws, in fact. However, even with those flaws, it demonstrated my point (that the GIF sucks) as far as I'm concerned. More optimal settings and all sorts of possibilities for alternatives should be explored. But, I will leave that to someone else because that is outside of the intended scope of my intent for this comment.

GyfCat reports the smallest file size between the both WebM and MP4. If the MP4 is smaller (or the WebM is smaller), it shows that in the textual information, regardless of video file presently playing. What's shown in the textual info is purely about the boasting in this case.

Still, that's definitely a file-size improvement over the GIF.

  GrimOpenDutchsmoushond.GIF:  8,714,192 bytes.
  GrimOpenDutchsmoushond.MP4:  1,137,305 bytes.
  GrimOpenDutchsmoushond.WebM:   378,363 bytes.

To me it doesn't seem that bad in this particular use case. Text is still perfectly readable and acceptable to me for its purpose in this context. Heck, you enlarged that image just to emphasize the visual artifacts around the characters. And at that, compression settings more tailored to this purpose could improve upon that. GyfCat uses "well-rounded" settings for their more general use case of converting any/all GIFs to video. Since this is for a very specific use case -- terminal screencasts -- you could probably improve on visual quality while maintaining improved (smaller) file size by tweaking the compression settings.

Honestly, if I'm reading terminal text like that example, I'd want it fullscreen on a small screen.

Eh... Admittedly, this is one scenario where GIFs might be acceptable if kept to small little (not long) screencasts for the terminal (simple, solid colors).

Still, movies win out in (1) smaller file size and (2) control for easy pausing, reversing, and skipping forward.

I hate it when I can't pause, go back, or go forward on instructional examples. Often I want to skip ahead to the relevant part, or pause so I can examine and play around more.

I want to be able to pause it easily so I can type on the computer keyboard I'm working on. I want to be able to scrub the timeline back and forth to easily replay something I might have missed instead of having to watch the whole thing over again. All too often a part might go by too fast, and with the GIF you need to wait for the whole thing to cycle again in order to catch that spot in the instruction that went too quickly. And then that spot goes by too quickly again, and you just want to pause the bloody thing, but you can't because it's a GIF. And so someone tacks on a bunch of extra JavaScript to create a player for the GIF in order to add that control. At this point in the story, file size and simplicity have both been thrown out the window. But, I digress...

The GIFs only serve a very limited purpose for VERY short screencasts so that it isn't an annoyance replay the whole thing again and again to catch something that was missed, etc.

Even with that example as small as it is, https://gfycat.com/AcidicTangibleChupacabra reduced it:

  GIF:  408,744 bytes.
  WebM: 344,978 bytes.
  MP4:  151,047 bytes.

That's a 2.7 to 1 compression ratio from the GIF to the MP4. And if you were specifically targeting "terminal recording" as a targeted scenario, you might be able to tailor the video compression settings for better quality and frame usage than GyfCat does.

I'd rather get the 147.5 kibibyte MP4 over the 399.2 kibibyte GIF on my mobile. Especially when I'm looking up something on my phone in concrete and metal building with poor signal and non-accessible Wi-Fi.

Using the HTML5 video tag, it serves WebM and MP4, so everybody should be happy.

There seems to be an abuse of GIFs when most people's browsers support MP4 or WebM just fine nowadays. If your browser doesn't support that, then you probably want to stick to simple text instructions for the terminal instructional example anyway.

A picture is perhaps worth at least a few hundred words...

http://i.imgur.com/VOlbMoy.png

icdiff seems to have the best defaults for me.

Yep! I forgot to mention the SRV records for some odd reason.

For a simple DNS-SD setup on unicast DNS, yes, it's basically a combination of PTR and TXT records. See http://www.dns-sd.org/serversetup.html for more information.

For example, in one network, multicast was completely disabled — no mDNS happening there — and so on our local (unicast) DNS server (BIND on a 'BSD), we added DNS-SD entries (PTR + TXT records) for printers (to a CUPS print server with Active Directory authentication) which gave all the iOS devices AirPrint capability to all the printers on the network (even AirPrint to old dot-matrix printers), entries for Apple TV/AirPlay, etc. Apple calls it "Wide-Area Bonjour" and it works with Macs and iOS devices just fine — it's Bonjour with no mDNS involved (more specifically, it's just DNS-SD with a unicast DNS server).

In a sense (and completely avoiding all of the details and particulars of the matter here, because that opens up a whole can of worms and a much longer discussion), mDNS is basically supposed to be the modern IETF Internet Standards branded replacement for the name service aspect of NetBIOS (and other local network name service protocols). See also http://lwn.net/Articles/252323/ ... Also, in a sense, Zeroconf (an umbrella term which includes IPv4 link local addressing + mDNS + DNS-SD) is replacement for SLP ... and SSDP & UPnP in a sense too (although they're not directly comparable concepts)... Read http://www.zeroconf.org/zeroconfandupnp.html for a better explanation on the UPnP side of things... (Again, using a really vague phrase "in a sense" here and glossing over the details and doing a hand-wavy gesture.) There are different matters of naming, service discovery, and higher layer protocols — and drawing lines of correlation aren't always direct nor as clear cut as some would like.

mDNS has utility in small networks for your average Joe. In these circumstances (1) it shouldn't be a problem, (2) it shouldn't flood (there's logarithmic decay and there actually has been thought put into this), (3) names are discoverable without relying on anything else, (4) and it does its job in being the mechanism for DNS-SD to piggyback on so that services can be discovered automatically. Being independent of other services working or not — e.g., DHCP, (unicast) DNS — means "it just works", even if other network services fail.

There is a beauty in the ability to just have all of your computers, printers, and peripheral devices otherwise plug into a switch (without any additional network service equipment — no router/dns/dhcp multi-combo device added in the mix) and that everything just ends up working. Link-local addressing takes care of IPs, mDNS takes care of names, and DNS-SD through mDNS takes care of automatic service advertisement — "Hey, I'm a printer, use me if ya wanna!"

Further, failure of proper mDNS implementation in a device (perhaps resulting in malformed packets or over-zealous flooding) is just a failure of that device's design, not mDNS per se.

Having said that, still... yes, I think the home/SOHO router/switch/access-point/dns/dhcp multi-combo device should handle DNS-SD registrations and DNS, and that mDNS shouldn't be needed.

In medium/large networks (those actually managed by someone putting on an IT hat), it's perhaps not as appropriate, but still shouldn't be a problem assuming everything is subnetted with mDNS in mind. Of course, name resolution is handled by unicast DNS, and with regard to service discovery in this context, someone manages the DNS-SD records on the unicast DNS server so you don't need mDNS to be the mule for DNS-SD to ride on in those circumstances.