TeX, LaTeX and Friends - 2015-07-12 (page 1 of 2)

« first day (1715 days earlier) ← previous day next day → last day (3233 days later) »

12:22 AM

@PauloCereda youtube.com/watch?v=SQhbSlQrLPQ :)

Title sequence for a new video series :)

9 hours later…

9:23 AM

@JosephWright while pulling out the name as a mandatory argument looks better if you just look at the provides_module call, overall I'm coming round to the single table argument version as in luatexbase, this seems to work OK and I'm not sure separating the name is better here:

luaotfload.module = {
    name          = "luaotfload-main",
    version       = 2.50003,
    date          = "2014/08/10",
    description   = "OpenType layout system.",
    author        = "Elie Roux & Hans Hagen",
    copyright     = "Elie Roux",
    license       = "GPL v2.0"
}


luatexbase.provides_module(luaotfload.module)

@DavidCarlisle I did say I welcomed feedback

@JosephWright yes :-)

1 hour later…

10:42 AM

@SeanAllred Oh it is Comfortaa indeed, I should've said that before. :)

@SeanAllred ooh <3

@SeanAllred oooooooooh

4 hours later…

2:17 PM

@clemens: have you ever used neon strings?

@PauloCereda no. I hadn't even heard about neon strings before. They look scary :)

3:07 PM

@clemens They surely do. :) A friend of mine has a bass with green strings. :)

4:00 PM

@SeanAllred: the suspense is killing me. :) Ducks are very curious. :)

In my continuing efforts to provide packages to CTAN that I myself never use, I have now created a package which allows one to embed film screenplay fragments into other documents. This package is based on a documentclass specifically for screenplays. Most the commands of that class apply to my package and some are remove (mainly having to do with the titlepage).

So here's the question: do I replicate the class documentation and then add a section describing the new functionality or simply point to the existing class documentation?

The latter strategy is what one might call the bigfoot strategy, and I tend to find it quite annoying. But on the other hand I don't want to spend much time on this. Comments?

@PauloCereda It's coming :)

But first, can anyone remind me which dimen controls the left margin? \dim_add:Nn \leftmargin {4em} doesn't seem to do the trick (trying to indent a block of text)

@AlanMunn Why duplicate the functionality in the first place?

Oh, wait – yeah, you can't exactly load two class files. My mistake.

4:16 PM

@SeanAllred Bingo. :)

@AlanMunn I would say to link to the existing documentation. Yeah, it's annoying, but it's The Right Thing™ – at least until the two codebases diverge for whatever reason.

@AlanMunn Maybe provide a 'proper' link (say, to texdoc.net or the CTAN mirror-pointer) and also the texdoc invocation.

@SeanAllred Yes, I tend to agree They are unlikely to diverge ever, since I doubt the class writer will ever modify the class, nor will I modify the package (except to fix bugs).

@AlanMunn I would document the differences, at any rate.

@SeanAllred Of course.

4:40 PM

@clemens Twitter suggesting me to follow LeFloid and dynamo Dresden ...
Can i get this to the list of *If you ever suggest crap like that again, i'll delete my account*?

@Johannes_B i tend to ignore any suggestions platforms like twitter or facebook make. They're not helpful but annoying!

@clemens Agreed. Just as annoying as our Library man.

5:13 PM

Jan 13 at 21:53, by Johannes_B

@ChristianHupfer No, i am not. I am just confused by the term template. We (the LaTeX community) need a clear statement on what a template is/ what a template is supposed to do.

There are a few other statements of me along those lines, datin back to september last year.

Nov 7 '14 at 18:08, by Johannes_B

It is the word template that bugs me in its complete meaning (or use?).

@Johannes_B Yep. As @barbarabeeton and I were talking about earlier, some sort of 'LaTeXicon' needs to be agreed upon and made prominently available.

It's something I think is certainly appropriate even for LaTeX-project.org or even tug.org

(Sorry, I have latex auto-correcting to LaTeX on this keyboard – saves quite a bit of time, normally :)

@Johannes_B @SeanAllred In a way every document class is a LaTeX template. But what we usually seem to mean when we say »template« is more of an example of usage…

@clemens Interesting – I've never personally seen the template/example mixup (but I can see how it could happen). I did try to make the distinction with that github project though

@clemens I stated exactly that a few months earlier, but coouldn't find the post.

And I agree with @PaulGessler on his definitions:

> To be sure, there are blurred lines here, and the terms are often used more-or-less interchangeably out there. But generally when I think template I think of something like a form letter, poster, or CV, where the custom styling is a large part of the code, and the content adjusted by the user is a relatively small portion.

> For examples, I think of a short sample file that users can build on, which is more commonly used when the content written/added by the user forms a larger part of the source code. Consider the "sample theses" often distributed with thesis/dissertation classes, such …

5:26 PM

@clemens @Johannes_B @SeanAllred At the risk of sounding like a broken record, I think attempts to legislate usage of this sort is not helpful. Suppose we do agree on a definition (and I like @PaulGessler 's version quoted above) we don't gain anything by spending time 'correcting' people's usage if there is any significant amount of variation among speakers on the use of the term.

@AlanMunn very true

@AlanMunn Certainly not, but it will allow us to communicate meaningfully with each other

And in the context of the site, correcting such usage can very quickly become super annoying.

@AlanMunn Yeah that would simply be obnoxious

@SeanAllred No, it doesn't. That's the point.

5:27 PM

@AlanMunn It doesn't?

@clemens @AlanMunn I wouldn't have ever considered such a confusion if clemens hadn't mentioned it

@SeanAllred Sorry, maybe I replied too fast. If by 'each other' you mean a few of us here, then maybe. But if by 'each other' you mean 'us and the OPs we interact with', then no.

@AlanMunn The few of us here :)

@SeanAllred Re »where the custom styling is a large part of the code« – actually reads to me like one should make a custom document class instead of cluttering the preamble with code

beamertemplate and package xtemplate?

@clemens Yes, but there's a point at which creating a document class for a single use becomes absurd and unwieldy

5:29 PM

@clemens The code is the scary part to a new user. Hide it. -> custom class/package

@SeanAllred which is we people like us do it and upload them to CTAN (cf tufte-book)

@Johannes_B I suppose as long as it doesn't get codified in CTAN

Well how about that

@SeanAllred You need basic knowledge to adjust to personal needs, no matter if in a documentclass predefined. And, lots of times we redefine the stuff, that comes with the class file.

My reasoning behind not uploading every template / document class to CTAN is, at this point (pre-l3), there's no good way to really and fully customize a class's behavior without diving into the code. If someone wanted to customize a template (which is often my experience with mentees), that would necessitate diving into a CTAN-released file – something we advise against (for good reason).

What does Microsoft Word define as a template? The basic stuff like headers/footers need to be made available using templates, if i understoof that correct on that one microsoft page.

5:32 PM

@Johannes_B Redefinitions – good point. But that still leaves at least some of the scary part, as you say, in the user-facing file.

@Johannes_B Word defines templates very broadly

@SeanAllred Sure, a template is something predefined, and to change it you need to know what you are doing.

They can be files ready for simple content or they can be fully-active files with automatic and manual fields, etc.

@Johannes_B I haven't used Word in a long time, but templates in Word are roughly similar to document classes in LaTeX, but they can of course have very specific stuff included in them (which is true for document classes too).

@SeanAllred it doesn't have to be CTAN. www.latextemplates.com could provide classes with a manual

@Johannes_B Not all the time :) I was just working with a friend yesterday on her template and, without any knowledge of TeX's macro processor, she was able to make a few adjustments in the (uncommented) code.

@clemens They already do, and it doesn't work.

e.g. latextemplates.com/template/asme

5:35 PM

@AlanMunn Yes, thinking about that, i would argue, that the documentclass is the template, as it sets the basic layout for the document. beamer for presentations, letter for letters reports for reports, or as some like to call it: theses

@Johannes_B @AlanMunn To be fair that is usually how I explain \documentclass to folks

@SeanAllred what do you mean with »it doesn't work«?

@clemens People aren't giving things proper manuals.

@SeanAllred Can you reply to specific messages? Makes it easier to follow.

@clemens Either that or they are inlining it entirely, which gets very messy (but isn't necessarily bad)

@Johannes_B The replies highlight the messages I intend to respond to.

5:38 PM

@SeanAllred that's the whole point of the discussion: the current situation isn't satisfying :)

@SeanAllred ? I have no clue what you mean by that

@SeanAllred who is they? and what is »inlining«?

Besides, there's no good way to reply to a message from a keyboard with external scripts (which my service provider is apparently unhappy with – the main site is almost completely unusable for me these days)

@Johannes_B Although I agree that functionally document classes are templates of a sort, users often also want an example document using the class, and this is what they think of as a template. But I think really people use meanings and this is why trying to decide on one will be not useful.

@clemens They: contributors of templates. Inlining:

%%%% PUT YOUR NAME HERE
\name{MY NAME HERE}

@AlanMunn an example of use?

@SeanAllred Code should to some extend be self-explanatory. if there is a command name and one title it should be obvious by just one line of comment what to input.

@SeanAllred <3

5:41 PM

@Johannes_B But certainly one can't be expected to know that \name exists in the first place

@Johannes_B Yes, as I mentioned yesterday, I don't call my thesis class a template, but I provide a document which uses the class with inline comments (as @SeanAllred bemoans :) ) and I call this a template, since I think that matches well with what lots of people expect the term to mean. Of course I have actual documentation as well.

Templates are fill in the blanks thingies.

2

@PauloCereda <3

@PauloCereda Fill in the blanks thingies.

@SeanAllred No, of course not. And every introduction to LaTeX would tell you it is author. That is why there is an example of use.

5:42 PM

@SeanAllred Ducks are good at explaning thingies. :)

@AlanMunn Done! :)

@Johannes_B Not for the ubiquitous res.cls

@Alan: I no speak English very good. :)

@PauloCereda Eu sabe quase tudo. :)

@AlanMunn Nomenclature needed :-)

@SeanAllred I consider this class obosolete, sorry.

@AlanMunn Eu não sabo. (isso dói na alma)

5:44 PM

@Johannes_B I do, too (all of them…)

That doesn't mean it's disappeared

@Johannes_B Well my users are quite well defined, so they quite quickly figure it out.

Consider the scenario: You want to want to write a report. A report needs a table of contents, headers, a bit of this and that. Search for template (or german Vorlage) gives you lots of examples of use, and every one of them make redefinitions to make it personal. Those bits are the bits that are badly coded. Without those, it would be report.cls with the stuff that is already defined. Maybe a few packages for extra functionality.

You can predefine a few, lets call them stylesets, to have the user choose from. If there is no suiting style of the five predefined styles for the header, there might be the realisation: Oh, it is not that common, i have to invest some own efforts. Of course, thinking about this two minutes longer will lead to L3 and (x)template or LDB.

@AlanMunn @SeanAllred @PauloCereda @clemens ^^^^^^

@Johannes_B Yep, was just reading. And that last portion is the rub – in the back of my mind, I feel like we're spending our energy on the wrong problems. Sure, this problem exists, but the real solution lies in L3 – the (x)template idea in general.

Surely we've seen a decline in the use of 2.09 in these last two decades (!!) since 2e's release. It will certainly be the same trend for L3.

user image

You gottta love some titles. :)

@SeanAllred Yes, but L3 is not ready yet. The template site gets 5000 hits every single day. That is redistibuting wrong code over and over again. Fixing this, will give some more time in the long run to spend on L3-dev.

5:54 PM

And as @JosephWright said (starred x5), we'll have a proper basis for what we call 'templates'

@Johannes_B I wouldn't call that a template. That's a recipe. :)

@SeanAllred where is that?

@Johannes_B Haha, @PauloCereda's starred image just pushed it down

@SeanAllred /dances

Hi, I think this should be moved from SO to TeX.SE
http://stackoverflow.com/questions/25998870/use-variable-in-sweave-code-chunk

5:56 PM

@Ho1 Should it? Sure, Sweave uses LaTeX, but it doesn't look specifically TeX-related

Actually, it is related to neither R, nor TeX, it is about literate programming, noweb. It was there for 9 month, with no answer. So, maybe it will be answered here.

@Ho1 Sorry, but (I wouldn't say) this site is about literate programming :( Certainly TeX has a long history with it, but it is distinct. (I've asked several literate programming questions on SO before, with success.)

@Ho1 Perhaps re-tag that question with literate-progamming

Jul 9 at 16:51, by Joseph Wright

Simple plan: I get the team to finish LaTeX3 then we have a proper basis for new templates :-)

Ok. I was just looking for an answer myself, and thought that this move could be a good idea.

@Johannes_B That's the one :)

@Ho1 You can always flag it for migration and an SO moderator will take a look at it and make the call.

6:01 PM

I re-tagged the question as you've said.

6:22 PM

@AlanMunn -- maybe it doesn't matter to anyone else, but here at ams we struggle with users who don't read the "readme" file distributed with "author packages". in each package, there are several parts, clearly marked: a class file, a template (xxx-template.tex), an example file, a readme, and sometimes more. many authors (especially new ones) tear stuff out of the example file, but usually not enough, or the wrong parts, leaving the production staff with a lot of extra cleanup work.

that is the major reason that i, at least, would like a clear statement of what we think a template is. (i offer the ams templates as evidence of what ams thinks a template is -- not an example file, and not a class file.)

2

@barbarabeeton I gave the authors of a Proceedings volume a simple sheet: "send your paper with the article class and don't add explicit formatting instructions”. My paper on horrors was born. ;-)

@barbarabeeton I think you and I have similar ideas on what a template file is. But my only point is that defining it doesn't solve your problem. Making it harder for users to do what they do would solve your problem.

6:37 PM

@AlanMunn -- have you got any suggestions (for making it harder for users to do counterproductive things)? the only pre-submission tactic i know is decent documentation, and we've worked very hard on that. we don't give feedback to authors of journal articles, unless something is so intractable that we tell them that if they don't fix it (according to our clear instructions), we'll have it rekeyed. (and people ask why journal subscriptions are so expensive.)

@barbarabeeton What kinds of things to they screw up?

@barbarabeeton I think the most stuff comes up when people trying to take a journal class and want to modify it to own needs. Same with reports/theses.

@AlanMunn -- changing margins (with geometry), using epsfig directly (our documentation -- and template! -- specifies graphicx), changing the main font from cm (or whatever is built in), ...

@Johannes_B -- changing a class to suit their own needs is okay -- as long as they don't submit that to the journal. in fact, we've tried to make the ams classes easy to modify in "reasonable" ways (ability to switch to/from italic for theorems, change size of bibliography text, that level of modification); our "requests" list has additional such elements.

@barbarabeeton These examples seem to have two different sources: changing margins and fonts may be as @Johannes_B suggests: people wanting to format things to their own liking. Using epsfig directly seems to be an issue of not being able to teach an old dog new tricks.

actually, since all ams classes are based on the three core classes, but not all books or journals look the same, such adaptable elements are needed for maintaining our own sanity.

6:47 PM

@barbarabeeton A somewhat draconian way to deal with things would be to simply not allow geometry (or other problematic packages) to be loaded, and to redefine e.g. psfig to do nothing etc., except with a [notforjournal] option. This wouldn't catch everything, but would be a kind of educational experience for authors...

@AlanMunn -- yes, we know that, and for an established "emeritus", make accommodations. (we still get an occasional \documentstyle or ams-tex file, but strongly encourage current ams-latex. and i don't mind saying that i'm terrified of the possible fallout from the core latex changes in tex live 2015.)

@barbarabeeton I encounter this where I am; there's someone who regularly teaches a LaTeX course in the summer, but it's filled with fairly old school advice.

@AlanMunn -- well, that would be a very nice addition to updated ams document classes. but, if you hadn't noticed, they haven't been overhauled since 2004, and are much in need of same. except that management thinks that other things are more urgent.

@barbarabeeton Yes, and if AMS won't put money/effort into maintaining and updating their classes, who will.

@AlanMunn -- some advice never goes out of style, fortunately. so it's important to concentrate on the things that can really get you in trouble.

6:52 PM

@barbarabeeton Yes, absolutely.

@AlanMunn -- you're preaching to the choir, but you know that. and if i sometimes get a bit testy, now you know why. i'm going to bother some folks in darmstadt with an idea i've come up with; we'll see what comes of it.

7:20 PM

@barbarabeeton Stefan is the admin of latex-community.org, which has teamed up with the template site in march. He will be a got one to poke as well.

@Johannes_B -- so now (assuming he's reading this), he'll know i'll be looking for him.

@barbarabeeton I had a look at the xxx-template.tex files. Personally I find them very clear and the comments instructive: there seems to be all information necessary. (What I can't really judge, though, is how inexperienced LaTeX users see this…) Maybe @AlanMunn as someone who teaches LaTeX (as I've learned just now) has some experience in that matter

@clemens Actually no, I don't teach LaTeX other than occasionally offering a mini semi-advanced workshop to students in my department.

@AlanMunn OK – but that's still more than me :)

@clemens But my experience with users and documentation, is that although good documentation helps, many users still don't read it. During my undergraduate years I wrote both software and documentation (for other systems) for a living.

7:32 PM

@SeanAllred surely xtemplate has no connection with "templates" as understood by latex templates site, apart from a re-use of the word, has it?

@AlanMunn I made that experience as a package author, too. (Although I can't judge on the »many« part…)

@DavidCarlisle Not in the slightest.

@DavidCarlisle But what a latextemplates-style template will look like in the future greatly depends on what the team settles on in terms of (x)templates.

@SeanAllred not sure how it got into the thread (but been out all afternoon, quickly scrolled over the history)

@SeanAllred as I say I don't see any connection really.

@DavidCarlisle I apologize for that, then :) I was meaning to say that instead of finding solutions for the current situation (2e), we should look to solidify the future situation (l3) so we can have a system in-place to handle latextemplates-style templates using the new l3 templates idea.

The terminology here is unfortunately very confusing :(

@SeanAllred Now this is an area where language can be legislated. If software designers have specific ideas in mind they should not try to repurpose existing language but create intuitive new terms that won't be confused. This is much easier than getting people to change language they already have.

7:38 PM

@barbarabeeton What was the estimated number of copies for a book by the AMS in the 80's? It's about my answer to tex.stackexchange.com/questions/254981/…

@AlanMunn Personally I was never going to attempt to effect global change :) Just something so we know what we're talking about as we're talking about it.

@SeanAllred @DavidCarlisle I understand that. But maybe the xtemplate terms need to be rethought so as to avoid this confusion.

@AlanMunn The problem is that xtemplate has a few fundamental flaws (that I'm not remembering off-hand). The idea as it stands isn't 'right' for l3, so coming up with a word necessitates a definition we don't have.

Can we replace the word template by any funny word, so we start things fresh instead of relying on confusing terminology? :)

@PauloCereda That's exactly the idea. But the word can't be so opaque as to not make some sense intuitively. So 'quack' is out, I think.

7:43 PM

@AlanMunn Oh that's no fun. :)

@AlanMunn the idea of the xtemplate system is templates for class writers, that is instead of having to write a class using low level tex coding you can basically just fill in a paramaterised template. But what's being discussed here are templates for document authors.

Perhaps author-templates vs. designer-templates?

To keep it simple.

Gotta love English – need a new word? Just shove some others together until it makes sense.

@SeanAllred I thought that was the definition of German

@SeanAllred Nah German is more interesting. Das Sehrkomplicatedauthorbasedtemplate.

@DavidCarlisle :) English and German go way back :)

@PauloCereda You know I actually thought that was real German for a few seconds.

7:47 PM

@SeanAllred It isn't?! /fades

<3

@PauloCereda Well 'sehr' is real, but I don't know about the rest of it :)

@egreg -- i don't have access to sales records, unfortunately, but earlier i did develop sales statistics for ams publications. books, in general, don't get enormous sales. when i was working on that, the all-time "best seller" was the russian-english dictionary, and that had sales of a bit over 10,000. for a "subject" book, the highest figure i remember was maybe 3,000; a print run of 3,000 would be taking a chance. it's lots easier these days to reprint, so the initial run is lower.

I think I will start coding everything in lua and call it from Latex. It is much easier than using Tex itself to program some logic inside Latex.  Here is my lua counter manager for example. Much easier than using Latex counters :)

\begin{document}
\reset
\add
\add
\sub
\add
\end{document}

It prints this
1
2
1
2

@Nasser Easy is in the eyes of the developer.

Here is the lua code:

% !TEX TS-program = lualatex
\documentclass{article}
\usepackage{luacode}
\usepackage{amsmath}
%------------------------
\begin{luacode}
local x = 0
function add()
    x = x +1
    tex.print(x)
end
function sub()
    x = x - 1
    tex.print(x)
end
function reset()
    x = 0
end
\end{luacode}
\newcommand\add[0]{ \directlua{add()}}%
\newcommand\sub[0]{ \directlua{sub()}}%
\newcommand\reset[0]{ \directlua{reset()}}%
%-------------------
\begin{document}
\reset
\add
\add
\sub

7:53 PM

@SeanAllred Of course designer-templates would then have to be parallel to designer-jeans, not better, just more expensive. :)

@Nasser in what way is that easier? It takes more code and involves cross-language communication which is always tricky. luatex is useful for many things but using it for that is bizarre:-)

@DavidCarlisle Once you set up the Lua code, and build the interface \command, the rest is easier. I find the logic and the code easier to read and understand and change.

@Nasser Also, why define \add \sub to insert a space before doing the arithmetic?

@SeanAllred But a name along those lines would probably help. Maybe 'design-template' instead.

@DavidCarlisle the above is just an example. Just did it in few minutes to see if I can save a counter inside lua. It is just meant for illustration of the idea.

7:56 PM

@Nasser also if using luatex use fontspec and unicode fonts, not classic tex ones.

\input expl3-generic \relax \ExplSyntaxOn % -*- expl3 -*-

\int_new:N \l_nasser_n_int
\cs_new:Nn \nasser_add:   { \int_incr:N \l_nasser_n_int \int_use:N \l_nasser_n_int }
\cs_new:Nn \nasser_sub:   { \int_decr:N \l_nasser_n_int \int_use:N \l_nasser_n_int }
\cs_new:Nn \nasser_reset: { \int_zero:N \l_nasser_n_int }

\nasser_add:
\nasser_add:
\nasser_sub:
\nasser_sub:
\nasser_add:
\nasser_reset:

\bye

@Nasser ^^^

Mine's shorter ;)

@SeanAllred thanks, but I found the lua code much easier to read and understand really. The code length is not important to me. It is ease of maintinance and understanding it. Lua code is much easier to read.

@SeanAllred should be \cs_new_protected:Nn, shouldn't it? :p

@Nasser I'm sure @DavidCarlisle could provide a Plain solution that's even easier to read.

@clemens meh, you're probably right

@SeanAllred i could:-)

7:59 PM

@clemens I honestly haven't gotten used to the idea of protected macros

It's a big gap in my TeX knowledge

@SeanAllred rule of thumb: if something's making assignments in the definition (set, incr, decr, …) the macro should be protected as assignments are not expandable

@clemens like this? (also @Nasser, @DavidCarlisle)

\newcount\nasser
\protected\def\add{\advance\nasser by  1\relax$\the\nasser$\relax}
\protected\def\sub{\advance\nasser by -1\relax$\the\nasser$\relax}
\protected\def\reset{\nasser=0\relax}

\add,
\add,
\add,
\sub,
\sub,
\reset
\sub

\bye

@barbarabeeton Thanks, I guessed a the right figure (3000). ;-) My point is that for 3000 copies, camera-ready typescript was much less expensive.

@clemens Oh wow, I think that idea of 'non-expandability' finally clicked

You can't really expand a macro that changes the state of TeX – once it's read, it just disappears

Right?

@Nasser Yes, then later on you will ask about performance, and we will be back at the starting point of this discussion. :)

8:05 PM

@PauloCereda performance it not a problem really, I could always buy the latest intel PC with 12 cores and 10 threads per core with 500 GB of ram if latex gets slow running :)

@AlanMunn That sounds good :)

@Nasser Won't help.

40

Q: Tips for choosing hardware for best LaTeX compile performance

Which things should one have in mind when one wants to buy new hardware optimized for best (pdf)latex performance? For example one might think the more cores on your cpu the better. But since latex is a linear thing there is no much space for parallelizing. So the number of cores seems not to be...

@Nasser ^^^

@egreg -- even so, ams switched long before that to selectric composers and varitype machines for "edited" books. only "author-prepared" material might use an actual typewriter. (i found a few in 1986-1987, and one from 1990 that looks like it was prepared on a selectric composer, in french, with accents added in ink, by hand. but after that, nearly everyone seems to have been using tex.)

@SeanAllred well by definition any macro can be expanded, but assignment doesn't happen during expansion but during evaluation (or as the texbook calls it, in the stomach)

@barbarabeeton The book in the question was by Cambridge University Press. Perhaps they switched later to “newer” technologies.

@DavidCarlisle Yeah, the assignment just continues on to TeX's stomach to be processed (just like a or \kern would, correct?)

8:15 PM

@Nasser but remember that unlike pdftex luatex is explicitly beta software and it has and will undergo breaking changes between releases. if you need to use unicode fonts or other things then this is quite likely a risk worth taking, but to print the value of a counter, probably not.

@SeanAllred more or less, yes

@DavidCarlisle Thanks :) Perhaps it's time to read the TeXbook again – I imagine it's something you keep learning from each time you read it.

New knowledge with experience and the like.

@Nasser :)

@DavidCarlisle Beta reduction from λ-calculus? :)

@SeanAllred just stare at the output from \def\foo{\relax}\def\bar{\foo}\edef\zz{\def\foo{\bar}} and all will be clear

@DavidCarlisle No console output – which should I run?

@egreg -- i did look at the question before answering. now i've looked the book up in mathscinet. cassels was a well known and respected author, but the book was in the series "london mathematical society student texts"; probably not expected to be a big seller, so it wouldn't have gotten the kid-glove treatment as far as production; i hope that editorial treatment was more attentive. anyhow, it got an excellent review (for content).

8:21 PM

@PauloCereda if only tex was a pure lambda calculus macro expansion system without these impure side effects like producing output, things would be so much simpler.

@PauloCereda I learned a new thing from yesterday's tour; perhaps also @JosephWright is interested . The mountain pass between Italy and Slovenia is very curious, because it connects a valley in Italy where water goes to the Black Sea (via the Danube), whereas the water in the Slovenian side arrives in Italy and the Adriatic Sea.

@DavidCarlisle :)

@SeanAllred tex '\def\foo{\relax}\def\bar{\foo}\edef\zz{\def\foo{\bar}} \show\zz\bye'

@DavidCarlisle Wouldn't that \edef immediately redefine \foo to be \bar? Would it further expand \bar back into \foo (where it still means \relax)?

@DavidCarlisle Ooohh

@egreg ooh we should definitely put a on-board camera on your motorbike. :) Are there waterfalls?

8:23 PM

@barbarabeeton I guess that 3000 copies for it is very optimistic and business is business also for CUP.

@DavidCarlisle I feel like I've realized something deep and important, but I don't see how it relates to the concept of not being expandable. Literally everything expanded, there.

@SeanAllred no the edef doesn't define \foo at all that's the point, \foo expands away before the assignment.

@DavidCarlisle Yes

@PauloCereda Sorry, no waterfalls.

@SeanAllred \relax did not expand, and nor did \def

8:24 PM

@DavidCarlisle Because they're primitives, I presume

@egreg I suppose the valley is very wide?

@DavidCarlisle Sorry if I'm a little thick-headed :(

@SeanAllred but if you had started with \protected\def\foo{\relax} then it would not have expanded and so \foo would be redefined if \zz was executed.

@DavidCarlisle Ooh, let me try that

@DavidCarlisle Is this akin to \exp_not:n?

Does being protected stop expansion at that macro?

etex '\protected\def\foo{\relax}\def\bar{\foo}\edef\zz{\def\foo{\bar}} \show\zz\bye'

8:26 PM

e.g., \protected\def\foo{…} – if \foo is encountered in an expansion context, it will not be expanded?

@SeanAllred that's just \noexpand :-) so that just works once and is used at the point of (non)expansion whereas a protected def is a flag on the token that makes it not expand in all expansion-only contexts

@SeanAllred yes exactly

@PauloCereda Not at all. It's quite narrow; there's a small lake and an interesting village born around some mines of lead and zinc back in the 14th century. It has a nice old church.

@DavidCarlisle I think I understand :) Now I'm just not sure why you would ever want to do this – it's of course necessary for some things, but I can't think of examples

None that make me say, Oh, yes, such a thing makes sense

@PauloCereda google.it/maps/place/33018+Cave+del+Predil+Udine/…

@SeanAllred you almost always want to do this, all user level macros want to be protected. latex2e does it with \protect and \declareRobustCommand which involves some elegant tex tricks but etex \protected works better and is far simpler

8:31 PM

@egreg Lovely!

@DavidCarlisle But why?

As you see, the river through the village “Cave del Predil” is a tributary of Gail, which in turn is a tributary of Drava. Near Tarvisio, at Kranjska Gora (Slovenia), a branch of the Sava river rises. The Drava and the Sava are among the most important tributaries of the Danube.

@DavidCarlisle I guess we are making progress with ltluatex

@SeanAllred think of putting a font change into a caption and watching what happens when that is written to the toc file if some kind of latex or etex protection is not used.

@SeanAllred \write :-)

8:32 PM

@JosephWright looks that way.

@DavidCarlisle I'm planning to get my talks/_TUGboats_ written this week so hope we can let people think a bit on the LuaTeX stuff (in any case probably we want to talk about it in person)

BTW, I'm guessing we'll want a 'live stream' TeX-sx chat room from TUG2015

@DavidCarlisle Wouldn't the font change be copied over into the toc file whether it's protected or not? (Unless the caption text is \edef'd somehow)

@SeanAllred or compare \protected\def\zz{\ifmmode math \else text\fi} and the same without the \protected and see which one works the way users expect....

Assuming sufficient power, I'll probably try to do some posts, plus like @StefanKottwitz did from India do a blog post per day

@SeanAllred no the definition of the macro would be randomly expanded and it would break, this is the classic "fragile command in a a moving argument" error of latex.

8:35 PM

2 mins ago, by Joseph Wright

@SeanAllred \write :-)

@DavidCarlisle Waitwait, does \def expand its argument once?

@SeanAllred no

@SeanAllred Really, texdoc texbytopic and search for \write

@DavidCarlisle both variations seem to do the same thing

@JosephWright doing that now

@SeanAllred try them at the start of a tabular cell:-)

8:37 PM

@DavidCarlisle :-)

@DavidCarlisle The definitions or the uses?

@DavidCarlisle Evil

@SeanAllred uses.

@JosephWright sorry

@SeanAllred Let me guess what you get from '\def\foo{\relax}\def\bar{\foo}\edef\zz{\def\foo{\bar}}: you're getting that \zz expands to \def\relax{\relax}

@DavidCarlisle Not your fault :-)

@DavidCarlisle I often wonder what my favourite e-TeX primitive is: \protected is up there

8:38 PM

@JosephWright I could have mentioned earlier where protection was needed:-)

@DavidCarlisle Yes, but some of the cases might be more confusing than helpful

@JosephWright It should have been in TeX from version 1. ;-)

@DavidCarlisle WHAT IS THIS DARK MAGIC

3

@egreg Yes

@SeanAllred welcome to my world

8:39 PM

@SeanAllred @DavidCarlisle Or mine

@DavidCarlisle I suggest we set l3galley as an exercise :-)

@DavidCarlisle Or is that one of the things I'm meant to update you on?

@SeanAllred tex reads ahead by expansion at the start of a table cell looking for \omit and without \protected the \ifmath is expanded to false always even in an array as the math has not been inserted yet. If \omit is not found that expansion is not undone so it _always takes the false branch even though it does then insert the halign template and set the cell in math mode. with \protected the token acts like \relax and stops the scan for \omit

@JosephWright both:-)

@DavidCarlisle Let me read that a few more times.

*mild epiphany*

@DavidCarlisle That does make total sense now, thanks :)

@JosephWright I'm still going to read that Expansion and \write section, though :)

@egreg Naturally :)

@JosephWright I guess it's not really our call but do you think plain luatex should load existing luatexbase (so Elie et al have to support luatexbase forever) or do you think we can push through an e@alloc based regime in plain. It wouldn't take much code hardly uses any of the rest of latex but might be too big a jump...

@DavidCarlisle There's a question about it on the site (@egreg and I have a minor difference of opinion about how to describe it)

@SeanAllred It's easy: \def is unexpandable, but \foo is (and \relax isn't). So \edef expands \foo until finding \relax; similarly it expands \bar to \foo and this one to \relax.

8:48 PM

@DavidCarlisle Plain LuaTeX already loads etex.src I think and is known unstable at present. I think it we offer the code and show it's OK Karl will take it.

@egreg Of course, that wasn't the hard part to understand :)

@SeanAllred that's why @egreg understood it.

@DavidCarlisle brutal

@DavidCarlisle Two-parter: luatex.ini loads pdftex.ini, the latter then loads etex.src

@SeanAllred but fair, I think.

8:49 PM

yesterday, by David Carlisle

@SeanAllred why should we be fair?

13

Q: Correct usage of \ifmmode

Seems that there is some subtlety in using \ifmmode that I am not aware of. For some reason, using: \newcommand{\hlcodeA}[1]{\ifmmode\else\small\fi\texttt{\hilight{#1}}}% is not quite identical to: \newcommand{\hlcodeB}[1]{\texttt{\hilight{\ifmmode\else\small\fi#1}}}% where I moved the \if...

math-mode amsmath align

14

Q: \ifmmode doesn't seem to work correctly inside an array environment

I’m struggling with getting \ifmmode to work correctly when used inside an array environment which is inside a displaymath environment. Here is a minimal example: \documentclass{article} \newcommand{\signal}[1]{% \ifmmode 1#1 \else 2$#1$ \fi } \begin{document} \begin{displaymath...

math-mode conditionals

@JosephWright yep. Then there's the context version of plain luatex, but I suppose we could forget that? (or to be honest why do we have two versions)

@DavidCarlisle Huh? the context version of plain luatex?

@SeanAllred All this talking about new words and the future, reminds me of Karl Pilkington:

Karl Pilkington: The Future [HD]

@JosephWright let me see if I can find it...

$ kpsewhich luatex-plain
/usr/local/texlive/2015/texmf-dist/tex/generic/context/luatex/luatex-plain.tex

@JosephWright ^^ I think not looked in detail at it recently

8:53 PM

@DavidCarlisle Not come across that before, but probably used only by the ConTeXt community I suspect

@DavidCarlisle Some interesting design decisions!

\newcount \lastallocatedattribute \lastallocatedattribute=255

\def\newattribute#1%
  {\global\advance\lastallocatedattribute 1
   \attributedef#1\lastallocatedattribute}

some questions on site, mostly asking about the difference (how I came to notice it) tex.stackexchange.com/search?q=luatex-plain

@JosephWright why start there?

@DavidCarlisle Dated ~2009 but then ConTeXt doesn't go to CTAN regularly so TL is not the best way to get their code anyway

@JosephWright I did try google to find an example of a plain tex file using luaotfload or luatexbase, but hard to search for

> They've gone mental with the L

:)

@Johannes_B :)

8:59 PM

@SeanAllred I had to laugh so loud first time i saw that moment :-)

@DavidCarlisle Getting a change in plain LuaTeX would make some of the issues go away :-)

@SeanAllred He's always mean to me; I'm never to him. :P

@egreg I look forward to witnessing a meeting in person

@egreg I've never understood it.

@JosephWright You should sell separate tickets.

@JosephWright ooh sword fight.

9:03 PM

@JosephWright We'll need to choose our seconds.

@egreg @DavidCarlisle will presumably choose Frank

@JosephWright I'll ask @barbarabeeton

@egreg Good plan

@egreg I'm sure she's very skilled with the sword

@JosephWright probably @egreg won't get to Darmstadt he'll be too busy watching the Ashes.

9:05 PM

@DavidCarlisle I am certainly looking forward to discussing it with @WillRobertson

@PauloCereda sword fight: youtube.com/watch?v=3I_Ds2ytz4o

@DavidCarlisle LOL

@SeanAllred No, that would be me.

@AlanMunn You fence?

@SeanAllred Yes.

9:07 PM

@AlanMunn :)

@AlanMunn maybe you should watch the above clip:-)

@SeanAllred Linguists are dangerous. :)

@PauloCereda No kidding – I only know two other proper linguists and both of them are skilled in swordplay – one of them with a katana

@DavidCarlisle If that's the case then @egreg has made a good choice: @barbarabeeton is the only one of the four who could go out right now and buy a gun (assuming she doesn't already own one.) :)

@AlanMunn Yay us :)

9:11 PM

@SeanAllred Well I'm Canadian actually...

@SeanAllred ooh

@SeanAllred 'murica!

@PauloCereda Your our graphics man: a suggestion for illustrating 'Unicodeland'?

@AlanMunn Well then @barbarabeeton and I will go to a range one day without you :)

@PauloCereda L3 logo will be on my slides :-)

@JosephWright oooh I do I do I do I do!

@JosephWright: would you like a beamer template? :)

9:13 PM

@SeanAllred Well I do live in the US, so I actually have the same possibilities. :)

@PauloCereda Just something for one slide: I go for a pretty light-weight apparence for overheads

anyone know if something changed in stackexchange sites? now when I login to Tex site, I find myself logged in to all the stackexchange sites I am registered in automatically, even without being asked like before? this started few days ago.

@Nasser Yes, that's a feature, not a bug :)

@Nasser Yes, this is a change that's been rolling out recently.

379

Q: Upcoming login changes (Stage 2 now LIVE)

As y'all know, our current flavor of "global authentication" leaves a few things to be desired. It's flaky, requires a page refresh, etc. etc. Wouldn't it be nice if you could just sign in once and be automatically logged in across the network? We are ready to roll out Stage 1 of Project "Make...

discussion featured login auto-login

9:14 PM

@PauloCereda Preview:

user image

@SeanAllred but this is not how it used to work. I do not like it. I prefer to login in myself.

@SeanAllred -- never owned a gun myself, but i did live with a bit of an "armory" for several years. got rather good with a flintlock, but if you miss the first time, it's all over. (unless you can simply vanish before the smoke clears.)

@Nasser Read the meta thread: STATUS BYDESIGN

> unless you can simply vanish before the smoke clears

that's my strategy

@SeanAllred Run away!

9:16 PM

@JosephWright Ah OK, I was thinking about something towards my Mary Poppins theme (which will be a success in our user meeting). Leave it to me, I will draw something for you. :)

Futurama Vines - Zoidberg Woop woop woop

@JosephWright ^^^

@PauloCereda twitter.com/UberFacts/status/620035359632695296

@Johannes_B ooooooooh

9:44 PM

@JosephWright: Done. Skype, email, here? :)

@PauloCereda Email :-)

@JosephWright Sent. Check if resolution is good.

@PauloCereda Looks good :-)

@JosephWright some thoughts on callbacks but it's a bit late perhaps better on latex-l or github but (a) can we unregister and set to false cf luatexbase.reset_callback(name, make_false) and (b) luaotfload uses a couple of user-defined callbacks (well it doesn't use them they are set to a dummy function, but they are there to be used..

@JosephWright You could use it as a background image. :) (it would require some lightning/transparency)

9:50 PM

@PauloCereda Like I said, I like traditional-ish slides so no backgrounds :-)

@JosephWright ... If it stays that way needs at least a very basic handler at least as in doctest, but actually reusing the handlers from the system callbacks probably shouldn't be that much more code. and if luaotfload needs it I'm wondering if after all it shouldn't be in the core....

@JosephWright ;)

@DavidCarlisle The reset business looks downright dangerous to me

@DavidCarlisle Will ponder for tomorrow

@JosephWright yes but... looking at luatex manual if you don't offer it you lose some functionality

@JosephWright: I could also put a little guy in the beginning of the path with a question ballon and �?.

@JosephWright: new version in progress. :)

9:57 PM

@JosephWright don't need reset (which removes all the allocated callbacks) just need a function that sets a callback to false if no callbacks are currently registered, or gives an error ortherwise

« first day (1715 days earlier) ← previous day next day → last day (3233 days later) »

Transcript for

Jul '1512

$TeX - LaTeX$ TeX, LaTeX and Friends

\begin{chat}

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

$TeX - LaTeX$

site design / logo © 2024 Stack Exchange Inc; legal

mobile