Hacker News Reader

TeX is an odd case, because it has been replaced in much of the world, with the almost-sole exception of mathematical and computer science academic writing. The vast majority of people doing typesetting of anything else (e.g. books, even at academic presses like MIT Press) will use a newer typesetting engine (one of the high-priced proprietary ones).

My guess is it just takes far too much implementation effort to replicate (or even significantly modify) something like TeX relative to the fairly small userbase of academic-paper-writers. The low-level typesetting-related parts in particular are pretty hairy and hard to get right, so experimentation tends to happen more in higher-level macro packages.

What are some of the high-priced proprietary ones?

Afaik, Adobe InDesign and Quark dominate the market, with InDesign being the de-facto standard. They're not as high-priced or numerous as they used to be, though; there were a bunch of sold-to-professionals-only packages in the 80s, but there seems to have been a convergence between desktop publishing and professional print prepress, so everyone just uses InDesign or Quark now.

I think FrameMaker would be another one.

You are only considering old technology versus new. There are old things that are much better than their newer counterparts. Take S-expressions for instance, or Lisp, or Tex.

These things worked amazingly well, yet managed to be replaced by lesser technologies.

Programming languages show clearly one of the reasons why this happens : the benefits might not be immediately obvious. You don't know what abstractions you are missing until you have taken the time to learn them and express your ideas with them.

In the case of Lisp, and Tex, there is also a huge problem of marketing.

To take your examples :

IPv6 is basically a necessity.

Windows to OS X is not really happenning seriously

SOAP and XML suck so bad that the benefits of REST and JSON are mostly oubvious

Chrome was well marketed, and purports to replace the obviously bad IE. It brings with it SPDY and Canvas.

"""There are old things that are much better than their newer counterparts. Take S-expressions for instance, or Lisp, or Tex."""

1) There are today tons of BETTER Lisp-like languages than 60's Lisp.

2) What are the "newer counterparts" that S-expressions are better from? Here's a few propositions for something better than old style S-expressions: http://www.dwheeler.com/readable/

And from a pragmatic standpoint, one can even argue that (more modern) C is even better than Lisp/S-expressions, because it has enabled orders of magnitude more used and useful software, include the whole of Unix/Windows/OS X/Linux, etc.

3) Putting TeX on this list makes it a vicious circle. You cannot explain why TeX has not been surpassed by something better by invoking TeX being better that anything newer.

This is not a thread for lisp/sexp discussions, but the better sexp miss the important point that:

define some(x y z) code goes here

should not be any different for the compiler, especially the way humans understand what the compiler does, than a simple list/array:

(1 2 3 4)

This way, you can mess with code the same way you do it with normal data structures, replace parts, add things etc.

The way they are writing it, "define" looks like a keyword, just like "if" and "for" is, in most of the languages, instead of a function, just like everything else in lisp is.

"We could accept all of what we have today, and take it for granted, and we'd live with it forever. But that would suck." The inverse of that is having everything change so quickly that we can't ever accomplish anything. It's all about striking that balance between improving tools and making the product or service.

Given how long Fortran has lasted in the scientific community, something is going to have to be remarkably better than TeX for people to switch. Old ways die slowly, especially in the sciences. As Kuhn's Structure of Scientific Revolutions observed, old ideas are usually only replaced when the old people die.

I do wonder how

Windows to OS X.

made it on that otherwise agreeable list.

Or IE to Chrome specifically. Speaking personally I find Chrome very feature-poor and much prefer Firefox, and nothing that's changed in recent releases from either looks likely to change that opinion.

What do you mean? OS X is a POSIX-certified UNIX, with a damn fine UI, and tons of both proprietary and OSS for it.

I can understand people liking Linux/Solaris etc more, but are there people that really believe Windows is better?

Yes.

It confuses the hell out of me that I'm defending Windows here - I'm an OSS enthusiast and run Linux whenever I can.

Thing is: Operating systems are at this point just a more complex choice of preference and taste. Pick anything recent (I'm not going to stand for Win 95/98/ME/XP), use common sense and best practices for the platform and you're golden.

No, OS X is not by default a better system. It might attract you as a BSD with a UI that you consider damn fine. Other people are free to disagree and like their choice of platform better, be it Windows, Linux, *BSD or whatnot.

I repeat: There's _zero_ ground for calling OS X generally superior (nor inferior!) in a war of operating systems, neither technically nor in any other way.

>Software evolves, and often dies when it doesn't.

That's what's confusing me about the piece. TeX preserves the ability to evolve (XeTeX, etc.) through stasis.

Or is it simply path dependence again?

It would be super tough to replace TeX for its intended users and uses.

There was a GRIM problem -- getting math typed for high quality work in math, science, engineering, and technology. TeX was a terrific solution. The problem has not much changed, and TeX remains a terrific solution.

Or the problem was to put black marks on white paper with math and English with high quality. Now we put black marks on a white background in a PDF file -- fine. How to do that was the problem, and TeX is a terrific solution. Or, the intended TeX users are for the intended uses concerned mostly with the math, just the math. For the math, we need the TeX math fonts and the AMS fonts and how TeX formats the math in 2D, but we don't need color, 3D, GUIs, popups, pull downs, rollovers, animation, hot links, text boxes, radio buttons, drag and drop, icons, etc.

Good TeX users have lots of macros for TeX, editor macros for TeX, papers in TeX, knowledge of TeX, documentation of TeX, skills with TeX, and journals that will accept papers in TeX and don't want to lose these or replace them.

For 'what you see is what you get' (WYSIWYG), 'what you see is all you've got', and for the intended users and uses, WYSIWYG is very much a step backwards (I would yell and scream bloody murder) and NOT wanted.

Similarly I care from not very much down to not at all for support of all of Unicode, color, graphics, animation, direct output of PDF, output of HTML/CSS, 3D, CAD features, etc.

For a lot more in functionality, I don't want to struggle with a LOT more in documentation, quality likely much lower than TeX directly from the hands of Knuth, years of bug fixes and version changes, time to acquire skills, time to write new macros, effort to struggle with journals that still want TeX instead of something new, etc.

TeX is one of my most important tools. To have me give up TeX will require twisting it out of my cold, dead fingers.

Net, there was grim problem, and it has not much changed. TeX was and remains a terrific solution. Done. That TeX won't slice bread is just irrelevant.