I'm likely gonna externalize them because numpy even though not being used or imported is causing issues that should have seemingly been solved by upgrading from 3.6 to 3.6.1, even tho I've 3.7 installed, what a farce
@vlg Don't get me wrong, but it is probably easier to complain about free software than maintaining or creating it. That is, you may want to refrain from words like farce in this context.
@marmot I've yet to use it, out of the box it doesn't work with tex, unless the pythontex package is doing something it shouldn't, or that windows is the problem. I'm complaining about a closed, solved issue, that I'd have to go thru hoops to just report. just inconvenient
@vlg Sure, some packages work more smoothly than others. I just want to encourage you to choose your wording in such a way that everyone, including the authors of packages (whom I highly respect), enjoy being in the chat. On the long run, you'll also benefit from this because it will increase your chances getting first-hand answers.
@Quintec So then not "literally plain text"... So add that environment to a small document, and then the actual text that causes the problem and post that.
Well, I don't remember writing that environment anymore, but I was pretty sure that was just for the title... I'm clearly wrong though
\documentclass{article}
\newenvironment{problem}[2][Problem]{\begin{trivlist}
\item[\hskip \labelsep {\bfseries #1}\hskip \labelsep {\bfseries #2.}]}{\end{trivlist}}
\begin{document}
\begin{problem}{1}
a.) How many different committees could be selected with no restrictions?\\
b.) How many different committees could be selected with exactly 3 women?\\
c.) How many different committees could be selected with at least one man?\\
d.) How many different committees could be selected with at least 3 men and at least 2 women?
@Quintec you just have a single paragraph of text with nothing to say the How should be aligned, so they are not aligned. (you should almost never enumerate by hand or use \\ in latex.
\documentclass{article}
\usepackage{enumerate}
\newenvironment{problem}[2][Problem]{\begin{trivlist}
\item[\hskip \labelsep {\bfseries #1}\hskip \labelsep {\bfseries #2.}]}{\end{trivlist}}
\begin{document}
\begin{problem}{1}(some intro text)
\begin{enumerate}[a.)]
\item How many different committees could be selected with no restrictions?
\item How many different committees could be selected with exactly 3 women?
\item How many different committees could be selected with at least one man?
How can I reduce the line spacing in between items?
Sounds great, thanks! I have thought I knew LaTeX for 2 years, but it turns out screenshotting some inline equations and creating a document are whole different things...
@marmot I wonder how much the code should be changed if the arrows.meta library is added (I can not find the changes to the modifications made by @Skillmon), shortcut https://tex.stackexchange.com/posts/453602/revisions
@manooooh If you're happy with the result, don't worry about it. Otherwise replace >=latex',-> by -{Latex}. (The syntax >=latex',-> is anyway unnecessarily complicated: ``>=latex'` says " all arrow ends are latex' and -> means "there is an arrow at the target", but since this is in a group, -latex' has the same effect.)
@manooooh I think so. Didn't try it out, though. If you want all arrows to be of that type in this picture, just do >={Latex} in the options of the tikzpicture.
@DavidCarlisle I've started looking over the test failure issue. I should probably arrange that l3build prints it's own version in the log in every run. However, by checking the install part of the logs, you can see that none of the failing ones update l3build. So they should all be using the current release, 2018-09-26 (makes sense date-wise too)
@manooooh I used >=Stealth in that specific answer though. The main usage difference between arrows and arrows.meta seems to be different names for arrow tips (and the latter has more predefined tips).
@JosephWright In some ways I wonder if we should test amsmath with normally configured formats rather than the base test format with its slightly odd font setup, the latter reduces tlg log diffs but the former is what people actually use. I suppose I should have two config setups and do both....
@DavidCarlisle At the moment I'm trying to pin down why I get more failures on my local system than I do on Travis-CI: I suspect it's a file in the full tree that isn't installed on the minimised system
@DavidCarlisle On Travis-CI and my VM, the failures are small and I think related to luaotfload changes. Ill probably fix Travis-CI then work out why I have a local issue ...
@DavidCarlisle They do test using the full tree, but I know what you mean
Hi all! Earlier today, an upvote on one of my answer postings earned me badge #1024 = 2^{10} on TeX.SE. Not a hugely significant event, for sure, but quite nice all the same. This event is also an occasion to announce that I’m “retiring” from active participation in this group. I might check in from time to time, especially if there’s a query about the selnolig package. But, for the most part, I will be turning to new and hopefully equally rewarding activities. Bye!
@DavidCarlisle My feeling is it's something to do with rebuilding the local formats (as we have tree searching enabled). I've zapped my ~/.texlive2018 and everything in ~/texmf, forced a format rebuild, etc.
@UlrikeFischer @JosephWright oh that's a point it always used to "install dependency base" but when I ran l3build check in amsmath just now it didn't make a format...
@UlrikeFischer Nothing has changed for l3build, the amsmath set up, etc. in this regard: before I rebuilt my local formats, I was getting the strange failures with an older checkout and older l3build: one that passed on Travis-CI
@JosephWright yes and no... I did l3build install; fmtutil-sys --all in base, first but I shouldn't have to do that should I (and I'm not sure now why that means I get OT1 fonts when testing in amsmath)
@JosephWright My current plan is to move all real issues to my site and close all issues on the lualatex site so that we know what really must be done. I want to make an update soon which includes the deva improvement. And then I want to work on my talk about tagpdf that I should make in less than three weeks. End of october we can start to think about the future.
@UlrikeFischer sure but if it's clear that the lualatex area is more or less frozen then that suggests a path that if anything does find a maintainer it should be moved out of there to wherever the new maintainer wants
@DavidCarlisle just leave lualibs and luaotfload for now so that I sort out the issues. For luamplib ask Dohyun Kim, and with the rest do want you want.
@UlrikeFischer "do what you want" == leave them there until anyone takes an interest in picking them up:-)
@JosephWright done, there is only one slot that matches with @egreg if that makes it difficult to fit everyone in, I may be able to do 12 on Monday or Tuesday (but I have meetings supposed to finish at 12 but usually run over)
@JosephWright oh well. 14.00 is normally fine for me with the exception of wednesday, next tuesday is unclear and now I have to do some shopping or we won't have something to eat.
@JosephWright engine=/ I'd not seen that:-) It would be better if that was the default if no engine was specified, then you could just do the natural kpsewhich latex.fmt
Good morning/afternoon/evening/night (depending on your time zone) to everybody. I’m writing this message to discuss an idea that come to my mind reading the following question:
In the inline math mode ($...$), if the formula is too long, LaTeX will try to break it on operators, e.g.
very long text followed by a very long equation like $a+b+c+d+e+f+g+h+i+j+k+l$ etc
may be rendered as
very long text followed
by a very long equation
like a+b+c+d+e+f+g+h+i+
j+k+l etc
...
TeX forbids line breaks at commas inside equation because of idioms like “f(a,b)” or “P=(x,y)”; anyway, there are other idioms (tipically, the enumeration of the elements of a set) where such line breaks are perfectly acceptable.
Why not to add, for example, to the mathtools package a declaration like \allowbreakatcommas (say), so that a user can type something like
Let $\allowbreakatcommas A = \{a, b, c, d, e, f\}$ be a set...
and permit line breaks at commas (with some penalty charged)?
@GuM possible, but in many ways it's similar to : or | where different uses require different handling and there the syntax uses different names so : or \colon or | or \lvert or mid etc.
@GuM so could just define \comma and use a\comma b\comma ...
@GuM Such cases should be handled carefully, like hyphenation: you generally don't want to have just one item before the break or just one after it. I prefer to get a overfull box and decide.
@GuM On the other hand, there are examples of long listings, I believe there are a few of them on the site. @DavidCarlisle and I have already discussed the problem.
@GuM actually there are newer possibilities eg I think in luatex you could make , a \mathbin` but locally set the spacing before a \mathbin to zero, but still allowing it to break, then you don't need active character definitions at all
@GuM he steals comments as just as he steals ticks.
@GuM You need some setup command anyway, so \setdesc{a,b,c} is as good.
@GuM With more work we can exclude the last two from being split across lines
\NewDocumentCommand{\setdesc}{m}
{
\{
\clist_set:Nn \l_tmpa_clist { #1 }
\clist_use:Nnnn \l_tmpa_clist
{ , } % between two
{ ,\allowbreak } % more than two
{ , } % between the last two
\}
}
In luatex you have a new possibility that does not involve active characters, you can declare , to be a mathbin (like +) so that line breaking is allowed and then set the mathord-mathbin spacing to zero so it gets no space before, like punctuation:
\documentclass{article}
\begin{document}
$
...
@UlrikeFischer for some reason @JosephWright and I would not be surprised, there doesn't appear to be any testing apart from the latex tests than we can see in luatex, don't know so much about context but...
Theoretically, would setting a box and putting an arg in then testing \ifdim\wd0=\z@ to see if arg is blank/empty be faster than a, say, \ifblank{arg}{T}{F} from etoolbox, or the plain \ifx- comparing with sth defined as a space or sth
@DavidCarlisle ;-). There must be some testing but it is naturally difficult if you explicitly assume that things will change anyway. But regarding testing, I see that Joseph blames babel?
@UlrikeFischer that's one possibility I don't think its clear yet and the tests pass here now (now I have installed latex from base), so the test suite isn't as isolated from the installed system as it should be
@JosephWright if I use my installed lualatex (installed from base) and simply do lualatex amsldoc-1 it uses TU encoded text (and would fail) but if I go l3build check it doesn't seem to make a format but lualatex.fmt appears in build/test from somewhere and the tests use OT1 encoded text (and pass)
@JosephWright actually there are so many differences in amsmath for unicode, that I wouldn't mind if it always used the installed format and do a on-off hit to update the tlg, as long as it's consistent whether it uses the installed one or the one used for base:-)
@DavidCarlisle I've not pinned it down 100% yet. I think it was a mismatch between files in the tree from babel (updated) and the format file (built in isolation), but I'm still not 100%
@JosephWright I looked a bit in the build folders. I guess that the formats are build in local and then moved to the test folder. In the local folder I have xelatex and a lualatex format. The time of the files looks ok - all from the time of the failing amsmath tests - but the fonttext.cfg is cleary the "normal" untweaked one.
@JosephWright sorry just saw that I wrote nonsense: the formats are older than the failing amsmath tests.
@DavidCarlisle, @UlrikeFischer Hmm, still haven't tracked it down. I'm sure it's to do with format building, but at the moment I'm not really keen on spending more time on it
@DavidCarlisle Are you busy on the 17th of next month?
I'm getting a strange result, whereby a counter gets increase by one more than it should if I use a) rather than b). Before trying to make a mwe, wanted to ask here. a) is \setbox0=\hbox{#1} \count0=\dimexpr\wd0/\dimexpr312pt\relax\relax
b) is \count0=\dimexpr\wd0/312\relax \makeatletter \divide\count0 by\p@ \makeatother
@JosephWright oh, oki! Is it advisable to write it in one way or another, or does it matter? For example, if we want to show the LaTeX code of $f'_x(0,0)$ I don't know if it is better to show $f_x'(0,0)$. Can you recommend something?
@JosephWright that was my first question! and as @boycott.se-yo' said, that was my second question. Yes, now I see why $f_x'$ is more legible than $f'_x$... but strangely I wrote it backwards... Maybe because the key combination I use to write ^ is closer to my fingers than _¯\_(ツ)_/¯
Oh, so now I just spent some minutes not understanding that -x**2 in MetaPost/MetaFun was interpreted as (-x)^2 and not the intended -(x^2). Hip hooray.
When the standard \leavemode is executed in vertical mode, the \unhbox primitive switches TeX to horizontal mode. The \unhbox token remains in the input stream at that point. This can be seen if we arrange to grab the first token in a paragraph using \everypar:
\def\temp#1{\def\temp{#1}\show\tem...
(I'm rep-capped already, so a second self-answered question seemed fair)
@JosephWright Is it OK to ask you a question here? If not, please just ignore the following. Question: is it possible to "shift up" a definition by one level? What I mean by this is something that makes a definition available just outside a group, but not global: \brgoup\bgroup\sgdef\X{1}\egroup\X\egroup should work but \brgoup\bgroup\sgdef\X{1}\egroup\egroup\X not because \X is now two levels outside.
Yes, a real shame. Probably eating ducks has serious side-effects. ;-) BTW, do you know the reason why someone may want to "protect" such helpers by using a @ in their name? (I always thought the @ is to prevent unintended abuse, but I can't see why this one is dangerous.)
What I mean to say is that with \loop you can make your compiler seem to be hibernating, which is harder with \@firstoftwo. All this seems to suggest that "locking these macros away from inexperienced users" is not the main reason.
@Skillmon actually it might have been \@loop in the kernel @@ prefixed commands are usually saved primitives, @ prefixed ones are more or less package interface commands and vowels-replaced-by-@ are lower level things, but the boundaries are blurred
@DavidCarlisle, you said I shouldn't put a \\ in my command to append rows to a table, why was it? something about the last row not being correct? My rows are correct, so it was sth else
@vlg you were adding \\ before you knew what was coming next which means that you could not have a \multicolumn in the first cell because your testing got in the way, you should do the tests before the \\
@DavidCarlisle Well, but you have to agree that there is a difference: in one version, there are two backslashes in the code and in the other these are not there.
@DavidCarlisle Uhm, I remember some questions I made on math.SE and these get views so fast (when using calculus, limits, etc. important tags)
@marmot ahá, the genius appears! How would you do to vote with your eyes closed without making mistakes pointing the cursor correctly? Only geniuses can do that!
@DavidCarlisle @JosephWright Yes, I know, @UlrikeFischer asked me to draw a diagram. After was done with that, I realized how much I love the duality net of M-theory. ;-)
@manooooh I guess that there's a lag between the “views counter” and the “votes counter”. Anyway, this one might have been set a record: closed and gone in two minutes.
@Shakaja without an example impossible to guess what you mean. Obviously they are not designed to mess up anything. by design they provide a normalised setup so that a minipage inside math or in a list has some standard context that does not inherit from outside, that is the whole point of the environment.
@DavidCarlisle On purely semantic grounds it seems that 'impossible to guess' can't be true. It's always possible to guess, just impossible to guess with any hope of being correct. #linguistmusings
@DavidCarlisle I'm using minipages mainly to have for example text and graphs next to each other. However when this includes theorems or ending on equations, it messes up my spacing. I tried various fixes for this (mostly from your answers to other similar questions), but cant get it to a point, where its consistent.
@JosephWright sorry, I wanted to answer in the afternoon but got side tracked ;-). Beside this: I'm the one importing all the dangerous new code from context so won't object about a bit etex ;-).
@marmot you did realize that the text of the marmot is a quote?
In the inline math mode ($...$), if the formula is too long, LaTeX will try to break it on operators, e.g.
very long text followed by a very long equation like $a+b+c+d+e+f+g+h+i+j+k+l$ etc
may be rendered as
very long text followed
by a very long equation
like a+b+c+d+e+f+g+h+i+
j+k+l etc
...
In luatex you have a new possibility that does not involve active characters, you can declare , to be a mathbin (like +) so that line breaking is allowed and then set the mathord-mathbin spacing to zero so it gets no space before, like punctuation:
\documentclass{article}
\begin{document}
$
...
