weeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee

@PauloCereda weeeeeeeeeeeeeeeeeeeeeee.....bang dinner!

@DavidCarlisle oh

You are mean :)

@PauloCereda I'm surprised they don't organize a match of ice duckey.

@egreg I think the underground world must have those... ahahah

@egreg duck curling. :)

@PauloCereda With ducks as the stones?

@egreg now I am confused. In retrospect, that might not be a good idea...

Oh man, I wish it was possible to turn off the GIF replay in the chat.

@wilx sorry

@PauloCereda Np, I just ad-blocked the image. :)

@PauloCereda just post lots of cricket scores to scroll it off the top

@DavidCarlisle Where's Psmith when we need him?

I had something very tasty for lunch. :-) Asian food. Crispy and tasty. But i cannot tell any more details. Sorry @Paulo :-(

@Johannes_B oh no!

You know what would be even cooler than @JosephWright's siunitx? One that you could write a value into, then flip a switch at the start of the paper, and it would print the value in those units. So when your prof tells you to redo everything in millimole instead of micromole you don't have to edit every value in the whole paper.

@Canageek You know what would even be cooler than that? A professo that let's you do your stuff.

@Johannes_B Well, yeah, but I understand the value of using the same units as the rest of your field.

@Canageek That is one point to look at it.

@Canageek probably wouldn't be that hard to add to siunitx, then you could use pound-mole instead of this si stuff:-)

@DavidCarlisle Even American chemists use SI.

@DavidCarlisle If I ever write a computer virus it will be a very sneaky one that will sit in memory and convert all Imperial measurements to metric and refuse to let you change them back.

@Canageek lol

@Canageek That's not so straightforward: squared units don't change linearly.

@egreg yes but if it's all just changing between different si prefixes so by some power of 10 it's probably not that hard to do accurately. Changing between metres per second squared and furlongs per square fortnight might be trickier.

@DavidCarlisle All doable but gets very tedious for me!

@JosephWright We need to get you a codemonkey. There must be an unemployed physics or math student around ;)

@JosephWright If you do ever make that extension, I'm totally giving my prof a paper in which all units are hogsheads and butts

@Canageek capacitor value in quadric mean solar days times square biots per square yard per stone?

Time to annoy @egreg : twitter.com/TeXtip/status/793541025717424128

@Canageek More professional? Per amor del cielo!

@egreg LOL

@egreg Well, compared to ( and ) without anything else?

@egreg latex.codecogs.com/… Which is better?

@Canageek That's even worse, yes. :(

@egreg OTOH, which is worse -- not being profound at all or pretending you are professional when you're not at all? :-/

@yo' I'd say \left( and \right) are fine. It makes LaTeX do the work for me, and makes it easier to tell which goes with which.

@Canageek Fractions are perhaps the only cases where \left and \right don't misbehave.

@egreg isn't that like, 90% of the use cases?

@Canageek Perhaps 10%

@egreg IMO, even around fractions the leftright size is too large, usually. (I know many people disagree with me in this. But.)

@yo' Sometimes; it depends on the whole formula: there's no hard and fast rule.

@egreg carve in stone!

@egreg but let's put it another way: I don't mind fraction sticking out of its delimiter a bit

anyway, time to go, the shop closes in 40 minutes.

@egreg Of all the problems I see with this, the \left and \right aren't the problem

latex.codecogs.com/…*%20%5Cright%29%20%5Cleft%7C%20H_%7Bso%7D%5Cright%7C%7E%5E%7B1%7D%5Cl‌​eft%28%5Cpi%2C%5Cpi%5E*%5Cright%29%5Cright%3E%7D%7BE_%7B%5Cpi%2C%5Cpi%5E*%7D%20-%‌​20E_%7Bn%2C%5Cpi%5E*%7D%7D%20%5Cright%7C%5E2%20f%5Cleft%28S_0%20%5Crightleftharpo‌​ons%20%5Cpi%2C%20%5Cpi%5E*%5Cright%29

@egreg errr [f\left(S_0 \rightleftharpoons T_1 \right) = \left| \frac{\left<~^{3}\left(n, \pi^* \right) \left| H_{so}\right|~^{1}\left(\pi,\pi^*\right)\right>}{E_{\pi,\pi^*} - E_{n,\pi^*}} \right|^2 f\left(S_0 \rightleftharpoons \pi, \pi^*\right)]

@Canageek Most of the \left and \right commands are wrong

\Bigl[f(S_0 \rightleftharpoons T_1) = \Bigl| \frac{\langle\,{}^{3}(n, \pi^*) \lvert H_{so}\rvert\,{}^{1}(\pi,\pi^*)\rangle}{E_{\pi,\pi^*} - E_{n,\pi^*}} \Bigr|^2 f(S_0 \rightleftharpoons \pi, \pi^*)\Bigr]

@egreg exactly! with a fraction sticking out :-)

@egreg But then they aren't taller then the stuff they encompass :D

@Canageek RIGHT!!!

@egreg So why don't you write a package that would do automatic sizing correctly?

@Canageek because:

@yo' because @egreg's no good at writing packages?

@DavidCarlisle yep; he's only lefting them.

@DavidCarlisle you cheat!

@yo' Yeah, but I'm betting you or @egreg's best guess is better then my best guess.

@yo' stupid language english i should stick to italian

@Canageek comes with practice; like you probably know, typography is an art, and (La)TeX is only one of it's best brushes; in art, all of practice, luck and talent count.

@yo' just like ms paint

@DavidCarlisle well, ever heard of pixel art? :-)

@yo' Bah. That just means we haven't done enough usability studies yet.

@Canageek who are you? a statistician? :D And also, it's an art, so the artist gives his opinion and judgement into it.

@yo' Chemist. My job is easier as whatever I do have a sample size of over 10^20, which makes life much easier ;)

@yo' But readability is a science, so you can eliminate all that messy opinion by finding whatever is the easiest to read. Then you establish that as the standard.

@Canageek oh no way! You have to keep that artistic part

@yo' Worked well for picking the font on highway signs.

@Canageek english text comic sans at 16pt

@Canageek oh not so different from my suggestion:-)

@DavidCarlisle I suspect that would lead to some issues with comprehension, though it is good for distance reading.

@Canageek that's a case where only the objective measures are important. But in typography, there's more into it. Note that you for instance choose font to suit the mood of the text. For instance, I like CM/LM for mathematical or engineering texts as the cut of the font is clear, very precise in some sense. I much more prefer e.g. Palatino for belletry and similar stuff, as the font has a nice touch of ... how to say ... it simply gives a warming feeling when you read text written in it.

Isn't the whole idea of LaTeX to, as much as possible, quantify and digitize the best practices in typography for you?

@Canageek with the emphasis on as much as possible. It can't choose the font, it can't tell you how big the graphics have to be, it can't tell you how large the delimiters should be in this case to make the formula look nice.

@Canageek The single fraction in the middle could have \left and \right; but there's the exponent which would land too high and, most important, the outer brackets: if you use \bigg in the middle (which is what \left and \right would do), also the outer ones should be \bigg, but they'd leave too much white space around f.

@yo' Yes, but this is only relevant for mathematical formula, so you've got a built in set of circumstances. You know precision and comprehension are important. You know that it will be in scientific papers. You know that lack of ambiguity is the single most important consideration.

@egreg \tfrac could be the solution here actually, but again, depends on the context of course...

@egreg Right, but it sounds like a logic tree could handle all of this. Also, how much of that is preference and how much would influence reading comprehension?

@Canageek not 100% sure about the last one being the single most important consideration.

@yo' Exactly.

@Canageek No, no and no. Now imagine you have a similar formula elsewhere in the paper that contains something which can't be enclosed in this size of delimiter. Now what do you prefer -- consistency between the formulae so that the reader at first glance considers them similar, or the technically most proper typesetting of each of the formulas separately?

(This is not the best example actually, but consider that you have to choose between \tfrac and \dfrac for each of them and for each of them separately, you would choose differently, but now you consider them being together, and you're stuck. You again have to decide how important is that they look similar.)

@Canageek With the due proportions, it happened to me the same as Knuth. I liked very much how my first published paper was printed. The second one went to the same typography, but they had bought a computer system instead of doing hand typesetting of math. The result was incredibly poor.

@yo' Probably have them all the same, or it could introduce confusion as to why they aren't all set the same.

@yo' Just appeared ;-)

@Canageek Note the word probably in your answer ;-)

@egreg ROFL

@egreg btw typo in your comment, missing )

@yo' In which case, I'd probably set an algorithm that sets them each separately, assigns some parameters, finds the formula that would suffer from being typset wrong, then reset it with all of the formula using that set of parameters, or some compromise between them.

@yo' Fixed, thanks

@yo' Right, you'd have to set up a case study, and sit a few hundred people down with eye trackers and a series of comprehension questions to know for sure.

@Canageek but the computer can't even check if the formulas are similar but completely irrelevant or not so similar but very relevant! Because the computer can't understand the paper contents as it's written in human language.

@Canageek The problem is that formulas have a meaning, the size isn't the only factor.

@egreg So do words, and we have a database on how to hyphenate, break, etc those.

@Canageek Formulas are two-dimensional, words are one-dimensional.

I'm still betting that you could create a set of rules that got it right 90, 95% of the time.

@egreg that's not the problem IMHO, it really only adds to the complexity. The problem is that formulas have a meaning, and the set of possible formulas is really infinite, whereas the set of words is moreorless finite.

@yo' Mine is slightly different from yours

\documentclass{article}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage[ngerman]{babel}

\usepackage{amsmath}

\begin{document}

\begin{align*}
\frac{E_{0-0}}{hc}
  &= C+\frac{E'(\upsilon'=0)}{hc} \\
  &= C+\frac{1}{hc}\biggl(
      \hbar\omega'_e\left(0+\frac{1}{2}\right)-
      \smash{\underbrace{\hbar\chi'\omega'_e\left(0+\frac{1}{2}\right)^{\!2}}_{\approx0}}
     \,\biggr)
     &&\mbox{\small
       \begin{tabular}{|l@{}}
       Vernachlässigung\\
       der Anharmonizität
       \end{tabular}%

@egreg yeah, I wasn't sure I prefer a tabular line over a big|, but it's probably better with teh tabular line. Also, this is exactly the case when I make the parentheses smaller...

We've got a TeX question on codereview.sx: codereview.stackexchange.com/q/145780/8953

@yo' That doesn't matter: Why would it? I want to show that these symbols are in these brackets, what is inside the brackets doesn't matter as long as I can clearly see what is in them, and clearly understand what is inside them.

@Canageek you're so wrong with "does not matter".

@MartinSchröder the example and tags have been pretty badly ruined by a mod there (I can't be bothered to join to complain)

@yo' Give me an example then. Way I see it, an integral is an integral and a superscript 2 is a superscript 2. Just make the sqrt and braces clear and you are good to go.

@Canageek sorry, I don't have time to bother. But if you want an example: I would possibly treat $A$ differently as an operator than as a number. I wouldn't like an operator $A$ to be in \tfrac for instance, with a number -- no problem.

@yo' If it is an operator it should have a ^ over it, which you can look for.

@Canageek your view on operators. Not everybody thinks so.

@yo' That is standard in chemistry. Chemistry is the best science, therefor others are wrong ;)

@Canageek sorry, I'm tired for such jokes.

@yo' I think you are taking this a lot further then I am. I just think there is an optimal placement of () that you should be able to infer from the formula, at least in the vast majority of cases.

@Canageek and I don't agree. That's all. Good night.