Have any one seen Dimension too large error when using breqn package? I keep getting these error due to large \begin{dmath}....\end{dmath}
! Dimension too large.
<recently read> \calc@B
l.393 \end{dmath*}
?
I posted a question on this tex.stackexchange.com/questions/219037/… but no answers. Thought to check here is someone seen it. may be there is a way to increase some configuration parameter or such to help with this.
@DavidCarlisle From c.t.t., what do you make of unicode.org/notes/tn28/UTN28-PlainTextMath.pdf as the 'MathML' (etc.) person here? I worry, as one often does, that a simple syntax of 'obvious' cases might run into issues with more complex situations. (I'm not even sure how they end up handling \sin(x) or similar.)
@Nasser no that means you really exceeded \maxdimen somewhere (5.7 metres) which is a hard limit in tex not configurable
@JosephWright oh that's Murray, it is the MS Word input form. It was (as I recall) originally drafted as part of TR25 but moved to its own TR28 as being somewhat specialised.
@Nasser you can always avoid the error by (manually or automatically) breaking things into smaller chunks, but you can't actually increase the maximum dimension as all tex registers are fixed sized registers holding at most 2^{30}sp
@DavidCarlisle I can't do manual edits, it is all automated code generation. I use breqn package. The test that generates long result, I just mark it down, and do not save its latex output. I can't edit by hand, since I run the program many times, and will lost any edits done. Maple is the worst, 99% of the long results come from Maple :)
@DavidCarlisle He's not so keen on MathML, is he (it makes TeX look very compact!)
@DavidCarlisle I wonder really how workable you think this is in general: is it a direction people in the TeX world should be thinking in (reusable input and so on)
@JosephWright I think it depends a lot on the input mechanism. If viewed as a texlike syntax for math (but with less explicit markup) it's a bit of a disaster as its grammar is internally very complex and if the obvious input doesn't do the right thing then parsing it and working out any other input is pretty much impossible. however viewed as a formalised linear input syntax for a math editor it seems to work well enough, in Word you just view it as extended "keyboard shortcuts" and ...
... just fix up any errors in the built up display as it appears.
Latex is amazing. The latex file now is about 1 millions lines long! and pdflatex compiles it to over 5,000 pages pdf file in only few minutes. That is why I use Latex and not word.
@JosephWright I arrived at a_b_c to be interpreted by “left associativity” as a^{b^c} which clearly means the author doesn't know what he's talking about.
@Nasser oh yeah, try even pressing PgDn in W@#$ with this long document :D
Barbara in an official tone :)
There's a little added variation this year. If a new member declares that an existing member was responsible for their joining, that (existing) member will receive a small thank-you token from TUG, and his/her name will be placed into a pool from which, at the end of 2015, one name will be selected as recipient of an original drawing by Duane Bibby, the artist whose illustrations grace the pages of the TeXbook, the LaTeX manual, and many other TeX-related venues. For more details, see this web site. (Don't use my name; I'm not eligible.) — barbara beetonyesterday
Hi guys, I've got a short question that I'm sure has been asked before: Is it possible to refer to an object (an equation in my case) before it is defined? Or is that just a dumb request? :P
@Danu \label and \ref always work via information input at the beginning of the document so forwards and backwards references work (or not work) the same way.
@Danu If forwards references work but back not then one of your "custom commands" has severely broken latex, and you should probably fix that first.
@Nasser conceptually the pdf has more stuff in it, as it includes all the font data and images but the dvi file does not, but it includes flate (zip) compression of its internal streams.....
if folks here buy books from amazon, watch out for book dates. Amazon put wrong dates to make the book appear new, when it is many years old. I noticed this many times before. Here is a book on Maple I just saw at amazon, it says it is published in 2012, when it is actually 1995 book. amazon.com/Maple-Handbook-Release/dp/1468402307/…
@Nasser You can't blame amazon for that: springer republished it in 2012 with a new isbn lookupbyisbn.com/Search/Book/978-1468402308/1 amazon is just no doubt using the publication date that the isbn returns.
@DavidCarlisle well, ok, but someone is trying to trick us. May be it is springer then. But this is not right. Old material, with just new date stamped on it.
A few weeks ago, the Bluesky website vanished. I have been unable to find any recent traces on the Internet of either Gordon Lee, who seems to have been the only person left at Bluesky (or, at any rate, its only interface with the public), or of its founder, Barry Smith.
Although Textures had lo...
If I produce a PDF with lilypond via the normal manner, a single page document comes out to 160K. If I run it through a script that uses pdftops -eps to turn it into eps, then uses epstopdf to turn it back into a PDF, I get a cropped PDF that is only 19K! Any idea why?
This is necessary because lilypond has a bug where it can't accurately detect its own bounding box to crop an image.
@AndrewCashner could be any one of thousands of reasons (it would be helpful iif I knew lilypond's output but I don't) it could be font embedding (pdffonts or acrobat font menu will tell you about that) it could simply be the compression level is higher or no compression at all in the lilypond output (pdf has zip style compression that can be set 0-9) or it could be something else or ...
@yo' @DavidCarlisle Thanks. I am including these images in a LaTeX document with \includegraphics, and there are a lot of them, so I need the file sizes to be small. This might be related: say I include 10 such files, and the text font for each is Linux Libertine O. In the pdffonts output for the LaTeX file made with pdflatex, it lists LinLibertineO 10 times. Is that normal?
@AndrewCashner in that case, you really want to un-embed the fonts somehow and embed the full font in the final document. However, with 20kB files, very likely only used subset of the fonts gets embedded.
@DavidCarlisle lilypond relies on Pango + Ghostscript for its font processing and printing. So perhaps its native settings for ghostcript are different than those used by epstopdf?
@AndrewCashner ah presumably your small images have the font subsetted, pdftex can sometimes but not always spot that two subset fonts are from the same font and merge them but sometimes it can't. So it can be that using the big images with full fonts results in a smaller final document because then the font can be loaded just once and subsetted to the combined set of characters used.
@AndrewCashner epstopdf is ghostscript as well as far as I know
@Johannes_B oh that was the other thing that my crystal ball suggested other than \mathsurround but I forgot to mention it. It had all the hallmarks of a user not saying what was actually happening.
@DavidCarlisle Here is the output of `pdffonts` for the cropped PDF. The only difference between this and the original PDF is that for the original, Emmentaler-16 has "no" under "uni": `name type emb sub uni object ID ------------------------------------ ----------------- --- --- --- --------- QYKHSR+Emmentaler-11 Type 1C yes yes no 14 0 YACNJQ+Emmentaler-16 Type 1C yes yes yes 10 0 EGIYSZ+Emmentaler-18 Type 1C yes yes no 12 0
@yo' Any idea how I could un-embed the fonts? I tried running lilypond with the EPS backend, but then epstopdf --noembed still tried to embed the fonts.
@DavidCarlisle Do I understand correctly that each PDF is subsetting a different portion of the font, so that for n included images there are n subsets of the font included in the final LaTeX PDF?
@AndrewCashner oh so they were already subsetted in the original
@AndrewCashner yes although as I say sometimes (depending on technicalities I don't really know) pdftex can detect that is happening and make one new susbet font with the union of the characters but in general if it can't do that it has to include the font with each image
@David @yo' Thanks. Supposedly lilypond's eps backend produces an EPS file without fonts embedded. So I tried including two such .eps files with \includegraphics. The result was the same; multiple embedded subsets of the same fonts. If you think this is a legitimate TeX question I'll post it on the main site.
@David Now I am trying ps2pdf -dEmbedAllFonts=false file.ps but it still gives the message "Cannot embed the font as it is too large, embedding a subset"
@AndrewCashner ah that's probably going to be a problem then, and may also be one of the underlying reasons why it can't merge the fonts anyway. Fonts are tricky things:-) You could try telling it not to embed the fonts at all when making the individual images and then (somehow) making sure it is available when all combined. Or give up, always a possibility:-)
@David I appreciated your literal take on a tiresome question. It's also more even-handed than the other answers, since the output on each program would be difficult to achieve on the other one.
Thanks!
I hadn't thought to look at unicode but tried Masroor's fix first. I only wanted to change the direction of E one time so I didn't use a \def. I had found the \kern-x statement somewhere sometime ago.
\textrm{\normalsize \rotatebox[origin=c]{180}{E}\kern-0.3em H\kern-0.4em P} \\
\textr...
Debugging nested TeX ifs are creepy: All I see is fi fi fi fi fi, look but you don't see, fi fi fi fi (to be read with a clown face and a 4 years old girl voice)
So I was reading Fixing LaTeX2e and it has several examples of the form
\RequirePackage{fixltx2e}
\documentclass...
What? I thought documentclass always went right at the top. Are there other situations I should require a package before documentclass?
So, inspired by Fixing LaTeX2e I've been wondering: Should I be loading fixltx2e and/or eTeX by default?
Typically I start my documents as
%Page layout here
\documentclass[letterpaper, 11pt]{article}
Should this be changed to
\RequirePackage{etex}
\RequirePackage{fixltx2e}
%Page layout he...
@ChristianHupfer The fifteenth letter of the Spanish alphabet is Ñ; in this case \csname foo\@Alph{15}\endcsname (or the same using the counter) would balk.
@egreg: Yes, I just thought about it... \@Alph is the backend of \Alph... I hope, the OP does not use Babel or does not need more than 14 boxes this way.
I am finding the idea of writing a program, that process input and generate Latex is amazing. This way, one can generate new updated documents, web pages, all in Latex, by running a program, which does all the work. This should work for any field, not just computer algebra. But if it is done from a computer algebra program, like I do, one can mix the computation with code generation.
@PauloCereda Hi Paulo. So I have a file Main.tex that \inputs one of two different versions of another file File1.tex or File2.tex. Right now I compile Main.tex with File1 and then change the name of the resulting PDF, then run it again with File2 and change the name. Could I automate this with arara?
@AlanMunn Hi Alan! :) Yes, I think we can automate it. :) Just a question: would you like a solution for the current TL version (3.0) or the new release (4.0)? :)
% This is the main file
\documentclass{article}
\begin{filecontents}{\jobname File1}
Hello world from File 1
\end{filecontents}
\begin{filecontents}{\jobname File2}
Hello world from File 2
\end{filecontents}
\begin{document}
This is the main file.
\input{\jobname File1}
%\input{\jobname File2}
\end{document}
@AlanMunn: first things first: arara 3.0 is way more restricted than my new version, so we will have to improvise. :) But it works. This is what I did so far:
@AlanMunn My pleasure. :) Note that this is a humble v3 solution. I'll come up with a v4 version (which is way more fluent and consistent) just for the record (I might release it in March). :)
I have a table that's a bit too wide with the normal font size and I'd like to shrink the font just enough for the table to fit the same margins as the rest of the page. Is there an easy way to do this?
@Gregory changing fontsize should be a last resort (there are dozens of question on site with this question) it's almost always better to adjust \tabcolsep or reformat the table a bit or use a predetermined size such as \small, but people anyway go \resizebox{\textwidth}{!}{\begin{tabular....\end{tabular}
@DavidCarlisle. \footnotesize is a touch smaller than needed, but it'll work fine. What's the difference between "changing font size" and "use a predetermined size"?
@Gregory some notion of document consistency and obeying the typographic style as laid out by the class. It's odd: no one suggests randomly scaling text in a paragraph to avoid overfull boxes in line breaking, but put the same text in a paragraph and you see it suggested most weeks.
Thanks!
I hadn't thought to look at unicode but tried Masroor's fix first. I only wanted to change the direction of E one time so I didn't use a \def. I had found the \kern-x statement somewhere sometime ago.
\textrm{\normalsize \rotatebox[origin=c]{180}{E}\kern-0.3em H\kern-0.4em P} \\
\textr...
