Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Always Bet on Text (2014)

graydon2.dreamwidth.org

Post reply on HN
Print view View original

241–250 of 349 posts

Re: Always Bet on Text (2014)

#241

Post by tannhaeuser »

dragonwriter wrote: post #94
crispyambulance wrote: post #82

Earlier quoted context omitted.

The funny thing about XML is that it was explicitly conceived to solve problems of representing, translating and transmitting component models and configuration. The sad thing is that it was abused, horribly, in the early naughts when people tried to take it too far. On the one hand it was rushed out without the proper tooling and you had humans typing raw xml without the right amount of support from intelligent auto…

> XML still survives, though. The tooling is good. If you type XAML or HTML in the right editors, it's serves the purpose competently and without drama. HTML isn't XML, though XML was inspired by HTML and HTML has an XML serialization. IIRC, some or all of the versions of XAML aren't, strictly, XML either, merely XML-derivedl0

> XML was inspired by HTML

No. HTML used to be "an application" of SGML. Meaning HTML brings a profile for SGML features (formerly, an SGML declaration determining things such as allowing tag inference and other minimization) plus a DTD grammar. Though HTML has also quirks for script/style elements, and URLs.

OTOH, XML, like SGML, is a markup meta-language and a proper subset of SGML (also with a fixed SGML declaration disallowing tag inference and almost all other minimization features). XML was introduced by W3C (the SGML "extended review board") as the markup meta-language for new vocabularies on the web going forward around 1997 to eventually replace HTML. While that hasn't happened, SVG and MathML have been specified using XML.

Details on (my pages) below.

[1]: http://sgmljs.net/blog/blog1701.html (the "TALK" slides)

[2]: https://www.youtube.com/watch?v=jy-b4jeJSas

Re: Always Bet on Text (2014)

#242

Post by boogies »

mrec wrote: post #198
aembleton wrote: post #193

Earlier quoted context omitted.

The Twitter logo in SVG format is only 687 bytes and can be as large as you like: https://upload.wikimedia.org/wikipedia/en/9/9f/Twitter_bird_...

My SVG is extremely rusty, but it doesn't look as if that xmlns:v="https://vecta.io/nano" serves any purpose besides advertising.

It's probably not the only removable part, the icon can be 414 bytes: https://github.com/edent/SuperTinyIcons#how-small

Re: Always Bet on Text (2014)

#243

Post by numlock86 »

doublerabbit wrote: post #234
numlock86 wrote: post #192

Earlier quoted context omitted.

Tell me in 723 characters or less what this "Twitter" looks like so I can make an exact copy of it. Text is ambiguous.

HN doesn't render ASCII or ANSI art correctly however both methods can serve as representation of Twitter in text and graphic less then ,~ ('v)__ (/ (``/ \__>' hjw ^^ Twitter Bird kindly borrowed from: https://www.asciiart.eu/animals/birds-land

That does not even remotely represent the Twitter logo, though. Shape and color are completely lost information, which are essentially the key properties that make the Twitter logo the Twitter logo.

Re: Always Bet on Text (2014)

#244

Post by hctaw »

> But let's hit the random button on wikipedia and pick a sentence, see if you can draw a picture to convey it, mm?

To be needlessly pedantic, my computer drew the image that conveyed this to me.

Less pedantically text is a medium of exchange for language, and a lossy one just like spoken word. I think there's a lot of power in its flexibility due to that lossiness. It's also one of its subtle weaknesses - we can read text from 5,000 years ago, but there's going to be much debate over understanding the text because of how much context has been lost to time.

Re: Always Bet on Text (2014)

#245

Post by danShumway »

doublerabbit wrote: post #234
numlock86 wrote: post #192

Earlier quoted context omitted.

Tell me in 723 characters or less what this "Twitter" looks like so I can make an exact copy of it. Text is ambiguous.

HN doesn't render ASCII or ANSI art correctly however both methods can serve as representation of Twitter in text and graphic less then ,~ ('v)__ (/ (``/ \__>' hjw ^^ Twitter Bird kindly borrowed from: https://www.asciiart.eu/animals/birds-land

This isn't text, you're using ASCII symbols as pixels in a grid to represent a raster image. This is literally just a raster image on an unconventional canvas.

If you treat a uniwidth character grid of characters as pixels, you have not converted an image into text, you have encoded text into an image. It's no different than using extended characters and font colors in a fixed width terminal window to make terminal art.

The fact that you use UTF-8 to encode something does not mean it will magically work in a screen reader, or that it will retain any of the advantages of pure text. I can encode any image or video file in a UTF-8 format and send it to you. Down this path lies calling everything text.

But if your "text" stops being readable if I disable monospace, or it stops working if I change the font, or it stops working if I change line wrapping, or it stops being communicable as soon as I try to read it out loud -- then it's not pure text. Using a UTF-8 character as a pixel does not change the fact that you're drawing an image out of pixels.

Re: Always Bet on Text (2014)

#246

Post by drewzero1 »

numlock86 wrote: post #208
drewzero1 wrote: post #199

Earlier quoted context omitted.

That sounds like a you problem. This site is one of a dwindling number that work flawlessly with a text-only browser, so if you view it in Links or similar (as I am right now) you can have a pure text experience without any of those pesky non-text technologies getting in the way!

> the "text-only browser"-argument That's a bit short-minded, don't you think? Or is this some kind of troll meme at this point? Anyway, I don't expect you to have some off-spec working TLS/TCP/IP/ARP/ETH implementation purely based on text, yet alone are connected to the internet via some text-only telegraphy station utilizing those things. Pretty sure your output device (read as: screen) has some sort of analog/dig…

~I'm not sure where you're going with this or how seriously you're taking it~, but if I have to I will dig out a couple of DECwriters and we can continue this discussion as a pure ASCII text conversation. (Just as soon as I find a null modem cable long enough to reach wherever you are.)

Edit: upon some rereading, I get that there was maybe more sarcasm in gp comment than I perceived at the time.

Re: Always Bet on Text (2014)

#248

Post by rhn_mk1 »

SubjectToChange wrote: post #237
numlock86 wrote: post #189

Earlier quoted context omitted.

By that same argument/logic you might as well argue cave paintings are superior over text. It traces back probably over 50000 years and is still universal today and doesn't even require understanding of a specific language - and it still forms the basis of most of society now. (art, painting, photographs, graffiti, ...)

Your reply is in text. Would you care to link a set of paintings to express your argument instead?

That's the point: take an arbitrary criterion, end up with an arbitrary answer.

Re: Always Bet on Text (2014)

#249

Post by mattnewton »

SubjectToChange wrote: post #237
numlock86 wrote: post #189

Earlier quoted context omitted.

By that same argument/logic you might as well argue cave paintings are superior over text. It traces back probably over 50000 years and is still universal today and doesn't even require understanding of a specific language - and it still forms the basis of most of society now. (art, painting, photographs, graffiti, ...)

Your reply is in text. Would you care to link a set of paintings to express your argument instead?

It’s also in ascii, is that superior to all other character sets, or an accident of history?

Re: Always Bet on Text (2014)

#250

Post by d0mine »

numlock86 wrote: post #93

Apples and oranges. The post states that the used Twitter logo PNG is 4000 bytes, while it's only 723. Try to express that logo in text, English language and in all its details with 723 characters, so you are able to reproduce a pixel perfect representation. Not even possible with 4000 characters anyway. Don't bother. Text will be ambiguous, unless you describe every single detail. If the target is to communicate "Tw…

Compare a book (a few megs) with a picture of the same size. What do you think can communicate more info? Do you have a single counter-example?
Post reply on HN