@CarLaTeX ^^^^^^ <3, thank you for having fun for a while
although with some complications I had a great time, it was hard for me to understand you but this is how everything works, persevering (even if I did not win any round!)
@DavidCarlisle : :-) Well, yeah that will work. But my files are in Dropbox and that means that the version history is lost when you move the files. This was a special case -- I do not use such file names in my normal workflow.
@egreg That works great? Thanks. Is there already a question where that solution is used? If not, I think it would be very useful question.
@PeterGrill that seems like no versioning. Moving a file shouldn't be the same as never had it. Start using a decent version control system like git and you don't have such problems.
@Skillmon Yes, Git has lots of advantages. Dropbox will still keep the version history of the older file, I just need to then remember that it was renamed. Plus Dropbox keeps a version history of almost every single change I make, automatically.
@JosephWright pressing is nothing. But what I would like to have in the feature is a "\dontnormalizeanything" command so that I can get a full log-file in the travis diff (it will fail naturally but it would be easier to see versions info of luaotfload).
@JosephWright Ah and one question: when I run the ctan target the pdf of the doc file ended in the maindir. Is this normal? And if yes how do you handle this? With .gitignore?
@JosephWright Good. So I didn't mess it up ;-). lualibs has a interesting build: one has to unpack the dtx, then one has to run mtxrun on two of the unpacked lua files to get the merged files, then one has to copy them away from the unpack dir and then one can run the next l3build target ;-).
@UlrikeFischer Hmm, tricky to support that in a generic set up ... one reason we've kept l3build itself as a 'raw' Lua file is that we'd otherwise have a nightmare in terms of extracting, etc.
@JosephWright I tried to output the fontloader version to the tlg but the line with fontloader-2018-09-19.lua disappeared. The docu says only that dates with slashes are removed. Is this intended?
I have a very long table that currently starts at the top of the page. In order to fit the whole table into the page, how do I shift it up 1cm, only if it starts at the top of the page?
@samcarter Yes, it is. It's just that my fellow marmots sometimes say that the crystal ball is too bright and sucks up too much power. So I only use it once in a while.
@marmot I don't get the power argument: shouldn't an underground burrow have easy access to underground power lines? Just be careful not to use your teeth to remove the insulation ...
@samcarter There are (luckily) not too many power lines at the elevation of our burrow. And marmots do not like to waste energy. This will accelerate the climate change, make the oceans rise faster and decrease our territory.
@marmot This is a valid argument for using the crystal ball only rarely. Although marmots are probably one of the last species to get in touch with the "ocean problem" :)
@marmot It is very easy to transform wine to water - next task: work on the back transformation :)
@Arszilla Don't worry, you already provided the answer by sharing the sharelatex link, there the normally open box is redefined with \renewcommand\qedsymbol{$\blacksquare$} In any case if you'd like more comprehensive list of symbols, try tug.ctan.org/info/symbols/comprehensive/symbols-a4.pdf
@marmot Every proof can be transformed into a one liner if a sufficiently large paper is supplied :) I once started to read Wiles proof but was lost somewhere in the necessary previous proofs to understand it.
@samcarter I am just wondering if Michael Atiyah's proof will be easier to understand... (It is actually amazing how long these guys stay active. Recently Yau came visiting us. He also still does things.)
Michael Atiyah is ~90 years old... He also tried to prove another conjecture but unfortunately the math community didn't support him, although he won an Abel Prize and Fields Medalla
For me 45 minutes is is very little time, so maybe he will say: the proof of this theorem is left as an exercise.
@ahorn why use longtable if you want it on one page?
@Arszilla avoid using \\ and if you do use it it's better to have no space before it (so at the end of the previous line) you could use \hfill $\blacksquare` but many theorem packages have a dedicated \qed command for that.
@ahorn you would nee \vspace* if that space is top of page (but using longtable seems wrong as I said.
@manooooh well only the usual reasons, that (a) getting the positioning right can be tricky (\hfill only works in simple cases for example) and using a command leaves choice of symbol to later if you use $\blacksquare$ then really it's going to be a black square but if you use \qed some publications may make it a rectangle or white or whatever that publications's convention is.
@manooooh sane reason that you should use \section{Foo} not {\large\textbf{Foo}}
@PeterGrill oh I never look at egreg's solutions:-), but files with # in them will bite elsewhere as it's a comment character in most shells so renaming the file might be good even if you were not using tex (but it wasn't an entirely serious suggestion not having the context of why the files are so named)
@samcarter -- wow! i hadn't heard about that. i'm rooting for sir michael! i met him in 1964, when i was working in the office of an ams summer institute in woods hole. (he was mostly hanging out with raoul bott, who didn't disappoint in his hungarianness -- brilliant and off-beat, which was certainly representative of all the other hungarians i'd met up to that time.) atiyah and bott were among the cast of characters responsible for the "woods hole theorem" in algebraic geometry. (cont'd)
(cont'd) unfortunately, the woods hole proceedings were never published, although i know the papers were written -- i typed and mimeographed most of them. anyhow, i can't think of anyone better to prove this hypothesis, and it would really demonstrate that a mathematician's work is not over when s/he turns 40.
@manooooh -- if you prefer the "native" code you might consider giving up latex and learing to deal with plain. (you'll miss an awful lot of texniques you've become accustomed to.) there are tradeoffs with everything. and if you've going to stick with latex, you should use it properly, otherwise you might get trapped when some package that you need makes different assumptions than you have.
@Krishna The right moment to try out a completly new luaotfload ;-). Well if something breaks tell me, then I can hopefully correct it before the upload.
@DavidCarlisle I had to set the version info into test2e to 1.07 to get travis working but then it worked. I'm just installing the experimental version to run the test with too. (I wasted an hour today trying to figure out what is wrong with one test until I realized that the gitignore prevented the upload of a core file ...)
@DavidCarlisle if you want to run test: the run files in texmf should be ok now, I'm currently correcting the build file for the ctan.zip.
@DavidCarlisle Sorry for the delay. No, if I cite something that isn't in the big it just leaves a ? in the document and prints an error in the log. If I've got \cite{mine} and \cite{Mine} in the same document and @article{Mine in the bib file it actually throws an error
@DavidCarlisle -- i say, tell @manooh the reasons and let him make the choice.
@manooooh -- tex (= plain tex) means that there are a few packages (e.g. amstex and eplain) but for the most part you really have to build your own superstructure. if what you're doing is sufficiently different from what existing latex document classes and packages are set up to do, it may be worth it, but it's a pretty long haul. better to learn to use latex properly.
@DavidCarlisle uhm, I will never use LaTeX/TeX for too complicated things... I mean, I don't need that. But I think to prevent some non-common errors I should not use external packages. My necessary packages are: amsmath, pgfplots, fontenc, geometry, babel, tasks
@manooooh I don't understand why you want to avoid packages (or at least that aim is rather strange if you then include massive package chain like pgfplots-tikz-pgf
@barbarabeeton wow, I think this is the same discussion to use C/C++ vs. Assembler. With Assembler we have more control on our decissions, but it's very tedious. Instead, C++ handles libraries but de have risks
@barbarabeeton hehe yup, but I don't think that's a package
@DavidCarlisle I'd have to check again and make a MWE. I just thought it was very odd that it would freak out so much at that type of typo. Also that it would treat A and a as KIND of the same thing. I'm use to as the same thing, or totally different things.
@manooooh -- you're correct: a document class is not a package, but it does many of the same things that a package does, and some (like memoir) make many packages unnecessary.
@barbarabeeton :) I learn a bit from two, but anyone tells me "DONT USE ASSEMBLER", " Use C++", and I prefer Assembler (even if I don't how to run a simple program)
Because I think it's more safer, there are less commands, it's more natural, I don't use "a single command of listings" or "a single command of that rotated page command", etc. If possible to do that stuff with the principal packages, I will do it
@manooooh much of the code that is in packages is only in packages and not in the main format because of the memory constraints on a 640K PC running emtex in 1993 when the latex2e format was designed. Whether its in an external package file or built in to the format is an implementation detail that should not concern someone writing a document.
@manooooh we see here all the time people posting "minimal" examples that load 100 conflicting and unused packages that have just been copied from a previous document (probably by a previous student) that isn't good either but there is no reason to avoid packages that you would use.
@DavidCarlisle do you mean that Knuth or the person who created LaTeX wanted to divide his main packages into others because of not spacing? Well, I see that all the packages are written by persons who aren't involved into other packages set up
@manooooh -- knuth considered plain tex just the basic necessities. for every particular type of document (taocp, the computers & typesetting series, his other books, articles for journals) he has created a tailored equivalent of a latex document class, containing everything that publication needs. not hundreds of "packages". he doesn't use (and doesn't like much of the philosophy of) latex.
@DavidCarlisle okay, that sounds fine for a person who works with LaTeX, but for a student like me I prefer to load the lowest packages if possible (my University told me that). I don't spend my life in LaTeX because I didn't grew up enough, I think. For example, even with you counterexamples I don't catch why de can't use $\hfill\square$ if, for me, works fine (altough I never write a formal proof)
@barbarabeeton do you think that he is from the "old school"?
@CarLaTeX nah, your draws were fine, I only think you needed colors
@manooooh -- if there's not enough space left on the line for the square, it will go the the next line -- at the beginning of the line, not the end. not what you expected.
@manooooh -- he has his own way of looking at things. and it essentially values simplicity, consistency, and economy. plus a very perceptive way of interpreting what he sees that is not necessarily how other people think.
@JosephWright How can I tell l3build to install a file from sourcefiledir in texmf/tex/... with target install but not put it in texmf/source when creating the ctan tds.zip?
@manooooh no, that is a good example, a package version has been tested over decades and deals with more cases \hfill\blacksquare will put the square on the right unless a line break happens at the \hfil in which case the space at beginning of line is discarded so it will be on the left.
@DavidCarlisle ok, I don't know how many times we have that kind of problems. One could solve it manually I think, but the most important thing is de are using a completely new package just for... prevent new lines!! This is not natural, takes more compiling time
@manooooh -- depends on the package. amsthm and amsmath go to great lengths to make sure the (empty) square is always at the right margin, and not separated onto a new page if the line it's supposed to end happens to be right at the bottom of the page. (at the end of a display it only works right if the display isn't numbered; that's a weakness that needs to be addressed, but doesn't happen very often.)
@manooooh which does the same as \hfill\qedsymbol in almost all cases, but takes care of the cases that a user typically doesn;t think of
@manooooh no if you have theorems and proofs you should be loading amsthm or a similar package anyway, If you just do it by hand the chances are the typestting will be poorer (the ams have a lot of experience of typesetting proofs)
@manooooh it is just the same as section headings. the latex format doesn't define \section you could just use some font change and by hand spacing but it is far better to load an external file (usually a class, but it could be a package) that defines suitable section heads.
@JosephWright I have no dtx or something like this in luaotfload. Only "direct" files which should go somewhere. I thought that this means that I don't need to set sourcefiles but then the target install doesn't do anything. And if I set it, the source directory in the tds zip gets filled with unnecessary files.
@JosephWright yes, but the build.lua there is outdated. Relative to which folders are the files searched? And what does "listed in both" mean if wildcards are involved? Must I use the same pattern?
@DavidCarlisle of course, but for the sake of simplicity, would you recommend extracting the code from a symbol or command from another package and copying it into the document itself? This avoids loading a package and speeding up times, although of course, it is possible that the commands used to create the command require the use of other packages... Is that possible and recommended?
@UlrikeFischer Shouldn't have to be the same pattern: the files are meant to be omitted by name. But I've not tried all of this with subdirectories, etc., so there may be issues. I'm looking, but if you can send an up-to-date build.lua as you use it, that would be handy
@JosephWright I'm just doing a few more systematic tests with dry-run. Is there a dry-run test for the ctan-zip? currently it always starts all the typesetting and this takes quite some time.
@manooooh no that would be a maintenance nightmare and vastly complicate for example converting the document to html or making accessible audio renderings or anything else a publisher might want to do with the document.
Please, can you express positive and negative opinions about this document clearly made with TeX / LaTeX? Is there something to improve?: https://arxiv.org/pdf/1610.09366.pdf
@DavidCarlisle ok wise man :)
Two things bother me: 1) that the author (or the established norms) indicate that the keywords are in a footnote; 2) endpoints are not added when the sentence ends with a math expression..
@manooooh note that (most) tex-to-xxx convertors don't actually read the tex code in packages. If you have \usepackage{amsthm} they have already tuned html equivalent code, so if you use begin{threorem} or \qed they generate suitable html. But if you just copy equivalent definitions or use lower level macros, they can not use the tuned html conversions they have to fall back on doing some basic low level conversion at the font and spacing level.
@JosephWright well when you look at: I need also an idea how to move filegraph.tex (in doc) to source and to remove filegraph.pdf from texmf\doc. That's only a pdf that I include in the main doc, not a standalone doc.
@DavidCarlisle I can not see it because constantly you propose commands on the site and you do not say "be careful with the conversion to html" or things like that
@DavidCarlisle buu!! Do you know what environment is "Current address" (last page)? It has tiny font and I often see it
@manooooh well if people ask for extended behaviour then some custom command may be necessary, but you asked if it is a good idea to copy definitions out of a package and use just bits of a package and there the answer is unequivocally no I would strongly recommend against that.
@manooooh I only looked at the first page, but it's at arxiv so presumably the tex source is available and you can check.
@DavidCarlisle I have already said, I do not publish articles, I am neither a scientist nor a mathematician nor have I received anything yet. I am a student. I never thought about using a conversion to html or using auidos or videos. Even under these conditions you do not recommend me to make this shortcut?
@UlrikeFischer At present, I can't get the docs to compile, which means I can't do install --dry-run --full. You'd not set typesetexe = "lualatex", but it's not just that. Is there anything I need to know?
@DavidCarlisle I do worry. In my university no student uses LaTeX, not even articles are published, only some notes that are posted on the university page but they use html (the source of the university page)
@manooooh amsart is style for ams articles so it is suitable for a math paper, it includes amsmath and lots of other things you'd need to load if using article, also built in suport for the the ams keywords etc.
@manooooh no journals use article, of course arxiv isn't exactly a journal and doesn't force a single house style but i would exeoct that they have a lot of amsart documents
@manooooh :sections are automatically numbered. But anyway given that he was approaching retirement age by the time latex was first developed you wouldn't expect that he'd be keeping up with tex best practice. he was probably concentrating on the math...
@DavidCarlisle The last step to publish a mathematical article would be to review the format. Nobody helped him? Maybe he was alone...
\noindent Ehresmann 1947: Introduced the notion of almost complex structure and showed that the 6-sphere admits an almost complex structure, but explicitly points out that he does not know whether it has a complex structure.\\ Hopf 1947: Proved that $S^4$ and $S^8$ do not admit almost complex structures.\\ Kirchhoff 1947: Uses octonions to construct an explicit almost complex structure on $S^6$.\\ Eckmann-Frohlicher and Ehresmann-Liberman 1951: Independently prove that Kirchhoff's almost complex
@DavidCarlisle I remind you that this article was implicitly not approved by the mathematical community. No one wants to refute it because he is a renowned mathematician
@DavidCarlisle he should have people by his side to help him with these tasks. I feel sorry that he has done all the work alone
What's the meaning of 99 in \begin{thebibliography}{99}?
I'm trying to understand the main idea of Atiyah's proof (https://arxiv.org/abs/1610.09366). Although there were discussions on MO year ago I couldn't find answers to my questions.
Consider the isomorphism on the 4th page:
"$KSp(\mathbb{R}^6) \rightarrow KR^{(7,1)}(pt) = \mathbb{Z}_2$"
My quest...
@manooooh -- this looks very like amsart.cls. the presentation of the first-page footnotes and the location of the addresses follow that style. the formatting of \curaddr has been modified; in ams journals, it would be strung out in a single line like the institution address above, but in ordinary roman text, not small caps. the absence of punctuation after math ending sentences is the preference of the author; this was not edited according to ams style, which would have added punctuation.
@DavidCarlisle I case you mean how to get a new fontloader working: with the new luaotfload version you simply install it (along with the new lualibs files, all is in the texmf). And better remove local luaotfload.conf (or comment all entries).
@manooooh -- the basic amsart convention is to include a journal logo in the upper left-hand corner. he is removing them, although there's a simpler, more latex-y way to do it (probably just to say \nologo, although i'm busy with something else so won't look it up). as @DavidCarlisle said, atiyah was already near retirement age when tex first saw the light of day. he probably used amstex first, and formed habits that are now out of date.
@DavidCarlisle I did put the lualibs files in the luaotfload repo, in the texmf so if you took this texmf you already have them. (luaotfload will imho work without it, but the font database is not recreated if they are missing).
@UlrikeFischer OK thanks, for now I'll leave it as it is, with teh path set up to include the texmf tree in your luaotfload area. I guess I should run the latex test suite.....
@manooooh -- this is almost certainly added by arxiv as part of their accession and management of documents. the font is not computer modern, so it was likely added separately to the pdf file.
@barbarabeeton the power of PDF files can adds clickable links to their documents haha, ok. I would have liked it to be a TeX/LaTeX functionality. Thanks again. Now finish your tasks! :)
@UlrikeFischer with test2e.tex as checked in to latex2e/support it doesn't segfault but if you change \ifnum\luatexversion<108 to \ifnum\luatexversion<208 (or \iftrue) so it does the "old" code it uses for luatex up to 1.07 then....
This is from l3build check -c config-TU in latex2e/base
@UlrikeFischer if you use a low level \font to load the same external font twice then classical tex show the first name in all future box logging but luatex uses the second name so lots of test files showed luatex-specific diffs, that font callback was suppsoed to just switch round the luatex font name logic an dit worked up as far as the luatex 1.07 but now segfaults, we could just not do that and add the extra .luatex.tlg files, but it would be good to know why it's breaking
You said "avoid using \\ and if you do use it it's better to have no space before it (so at the end of the previous line) you could use \hfill $\blacksquare` but many theorem packages have a dedicated \qed command for that."
Sorry for my late response as I fell asleep
My code was:
\subsection*{Solution} True. If $a \mid b$, then $b = ax$ for some $x \in \mathbb{Z}$, Then $bc = axc$. \\Thus $a \mid axc = bc$. If $a \mid c$ then $c = ay$ for some $y \in \mathbb{Z}$. Then $bc = ayb$.
...\testb a
...\testb a
...\testb a
...\penalty 10000
...\glue(\parfillskip) 0.0 plus 1.0fil
...\glue(\rightskip) 0.0
..\glue(\parskip) 0.0 plus 1.0
..\glue(\baselineskip) 5.05556
..\hbox(6.94444+0.0)x469.75499, glue set 433.08827fil
...\hbox(0.0+0.0)x20.0
...\testb b
...\testb b
...\testb b
...\penalty 10000
function injections.setmove(current,factor,rlmode,x,injection)
local dx=factor*x
line 15486 of /home/luaotfload/texmf/tex/luatex/luaotfload/fontloader-2018-09-19.lua
factor is nil (but it may be fault of my lua callback, although it used to work)
LaTeX2e <2018-04-01> pre-release-6
luaotfload | main : initialization completed in 3.350 seconds (/home/davidc/texmf/tex/latex/base/article.cls
Document Class: article 2014/09/29 v1.4h Standard LaTeX document class
(/home/davidc/texmf/tex/latex/base/size10.clo)) (./tu-tl2e7.aux)
warning (node filter): error: ...ad/texmf/tex/luatex/luaotfload/fontloader-201
8-09-19.lua:15485: attempt to perform arithmetic on a nil value (local 'factor'
)
Segmentation fault (core dumped)
here is a self standing tex file, extracted from the test suite setup
% Test file for nfss testing the style file for use
% of the DC fonts with NFSS.
%
% RmS 92/08/13, last changed 94/05/17 for 2e
\directlua{
local original_fontloader=font.read_tfm
if (luatexbase==nil) then
callback.register('define_font',latexDefineFont)
else
if(luatexbase.in_callback==nil) then
error('update luatexbase')
else
local cbl=luatexbase.callback_descriptions('define_font')
if(cbl[1]\string~=nil) then
original_fontloader=luatexbase.remove_from_callback('define_font',cbl[1])
