@UlrikeFischer Yes, but it still seems to work. Given that xparse is the recommended way to create document level macros for stuff defined using the expl3 syntax, as I understand it, I take it this certainly should work. My example obviously doesn't fit that description, but other cases without arguments obviously will.

@JosephWright I will have to trust you on that as I have no idea what anybody is talking about and I don't remember xparse's documentation explaining it either.

@JosephWright So is @egreg 's method the right way to do it? Is there a right way? Or is this just not supported in L3?

@JosephWright I have lots of cases where I define a new command which takes no arguments. Are they all going to break?

\NewDocumentCommand \cyfres {}
{
  \l_cfr_cyfres_tl
}

If I'm not meant to use \NewDocumentCommand when there are no arguments, what am I meant to use?

\newcommand \cyfres
{
  \l_cfr_cyfres_tl
}

??

@cfr I missed the beginning of this discussion but I wouldn't have expected the empty argument case to be a problem!

@VincentVerheyen templatesize, not \templatesize.

@cfr What's happening is that fancyref#2labelprefix is used inside a name construct (\csname ... \endcsname) by fancyref. This therefore has to expand to 'pure' text, and is not 'really' a document command, more a variable. (This entire area is tricky.)

@VincentVerheyen Don't know where that comes from, an MWE would probably be a good idea.

@cfr In the past we've used a 'shortcut' for m-type arguments (which includes no arguments at all) which means that you've ended up with a macro that will 'work' inside a csname, but that's accidental: happens because \csname doesn't respect \protected (for various reasonable reasons)

@cfr As you want something that expands to pure text, the 'xparse way' would be \DeclareExpandableDocumentCommand. However, there probably is a case here for us to consider: I'll raise with the team as a whole.

@cfr Commands with no arguments are always a bit odd: are the akin to functions or akin to variables in other languages ...

@cfr Mail to the team sent: I'll let you know what the wider feeling is

@PauloCereda Now that I'll probably become a TUG member, may I run the electoral campaign for @egreg? (Even if I think only Italians can understand the paraphrasis).

@egreg They won. :(

@CarLaTeX :)

@VincentVerheyen Ah right, neither one of us has read the manual properly: templatesize is an option to \includepdf, not to the package.

@PauloCereda yipee!

@DavidCarlisle o.O

@PauloCereda I'm interista, I can understand you!

@CarLaTeX <3

@DavidCarlisle ^^ you are mean

@JosephWright Well, these things really ought to be \...Expandable...

@cfr @JosephWright or \tl_set_eq \cyfres \l_cfr_cyfres_tl

@CarLaTeX Antonio La Quaglia!

@egreg Of course! An Italian immediately recognize him!

@yo' Well yes: the entire question of variables at the document level is yet to be addressed

@yo' I've raised with the team the question of whether we might make this \DeclareDocumentCommand*: there are more use cases for expandable commands than we first perhaps thought ....

@JosephWright well theres declareexpandabledocumentcommand

And also, I'd persoannly favour the \tl_set_eq solution, but then the question it whether you shall \tl_new on something that's not in expl3 convention

(sorry for the typos, i'm on smartphone)

So maybe something like \tl_set_pubilc:NV

@egreg Ma era La Trippa, La Quaglia era il capostazione di Piovarolo! :):):)

@CarLaTeX Vero! C'era Destinazione Piovarolo un paio di sere fa su non so quale canale.

@egreg :):):)

@yo' That's allowable (we have somewhere to do it), but the thing is that's a code solution so suggests it should be wrapped up in a way that doesn't need expl3 syntax

@yo' Of course, but the reason for the rather long name (and lack of New, etc.) was that we thought this would be very rare. It is, but perhaps not as much as we originally imagined.

@JosephWright the lack of New makes sense, and I'd stick to the longish name personally. Why not?

@yo' Well Declare is really a fall back: for creating a document-level syntax \NewDocumentCommand is recommended as it catches any clashes (cf. \newcommand vs \def)

@yo' On 'how long should a name be' there is of course no hard line, but very long ~ discouraged was my feeling

Unicode appears to have imaginary unit i character. In some fonts it looks like lower case i outline with white interior. Is there a macro for such i look?

@wilx Please! Don't inflict such horrors to your readers! ;-)

@egreg Haha. Is it that bad? :)

@wilx it is Mathematica syntax so introduced post 1700 so too modern for egreg (but yes it is pretty horrible)

@wilx It's something introduced by Wolfram. They like to be innovative.

@egreg ooh Wolfram

@wilx it's \mitBbbi in unicode-math or the stix package for pdftex or just type it directly as ⅈ

@DavidCarlisle OK.

I can imagine it could be useful if you are dealing with electric currents equations and impedance.

@wilx there was discussion here just a day or so ago about the matching d, I'll find it...

@JosephWright well, certainly for cases like "publishing" tls you don't need New

@wilx me barbara and mikep starting around

@DavidCarlisle Hah. I did not know there was a ⅆ even. I just checked, it appears we use upright d and italic x. :)

I think you could use 💭 for imagnary

@wilx don't tell @egreg :D

@DavidCarlisle Haha.

@JosephWright Thanks. I would particularly like to know if my more standard cases are wrong (e.g. the \cyfres case) because I have a lot of these and it is going to be quite bad if they suddenly break. If this is incorrect, I would like to know now so I can start changing them before they do break. I did them this way because I thought this was what I was meant to do to provide access to a L3 variable or function. (The fancyref case is obviously different. However, \newcommand works fine.)

@yo' How would this work?

In the real case, \l_cfr_cyfres_tl is set using a key-value syntax. Of course, I could reset \cyfres to the new value each time the key is processed, but that would be error prone. In that case, I would be better to just use \cyfres internally and drop \l_cfr_cyfres_tl altogether in order to reduce the possibility of my introducing errors.

\documentclass{article}
\usepackage{xparse}
\ExplSyntaxOn
\tl_new:N \l_cfr_cyfres_tl
\tl_set:Nn \l_cfr_cyfres_tl {first one}
\tl_new:N \cyfres
\tl_set_eq:NN \cyfres \l_cfr_cyfres_tl
\NewDocumentCommand\setme {m}{\tl_set:Nn \l_cfr_cyfres_tl { #1}}
\ExplSyntaxOff
\begin{document}
\setme{this one}
\cyfres
\setme{that one}
\cyfres
\end{document}

@cfr well, then really \DeclareExpendableDocumentCommand \myfres {} { \tl_use:N \l_cfr_cyfres_tl }

@yo' But I don't want a \Declare! I want to know if there's a name clash. But I don't really see why \NewDocumentCommand shouldn't work. The xparse documentation never mentions a requirement to have arguments.

@cfr True, and it's not that they have to have arguments, it's that there is an assumption that document commands are in general protected from expansion: csnames are a bit odd in that regard

@cfr If you'd tried something with an optional argument, for example, it would always have complained

@JosephWright So is my \cyfres case wrong or not?

@cfr Like I said, we've really not dealt yet with document-level variables, which is what this is: personally at present I'd go with \newcommand for such a case as it is expandable

@cfr Really one for the LaTeX-L list as others (Frank in particular) will have useful views

@JosephWright I'm much more concerned right now about the standard cases. (Standard from my perspective.) As I understood the discussion, the claim was that I shouldn't be using \NewDocumentCommand to access, say, \l_cfr_cyfres_tl. But there is no \NewExpandableDocumentCommand.

@cfr unless the command is Expandable, it's protected so it won't work in certain contexts. And the question whether \NewExpandableDocumentCommand is necessary is independent of this. I feel its usage is so limited that it's not necessary; you can simulate its behaviour after all

@cfr Yes, that is all true: we don't have \NewExpandableDocumentCommand as when we were setting this up we took the view that the common case is that protected status makes sense

@cfr This is where xparse is about trying to explore ideas: there's not always a immediate answer as they depend on the code getting used!

@yo' @JosephWright Do I need it to be expandable?

@cfr yes as it's used in csname

@cfr Certainly in your original example 'yes'

Also, you can use, in theory \cs_new:Npn \cyfres { \tl_use:N \l_cfr_cyfres_tl }

@DavidCarlisle Should we add an \ifincsname check somewhere to pick up this sort of thing?

@DavidCarlisle No it isn't. It is just used to access the value of \l_cfr_cyfres_tl in the document.

@cfr well, obviously, as the protected version leads to errors :)

@cfr That's more tricky: 'access the value' as in 'typeset' doesn't need to be expandable

@cfr your first example was used in csname (by fancyhdr, that's where it died)

@yo' No, it doesn't. The original case with fancyref did. This doesn't.

@cfr My take is that where you are representing 'simple' text then having the set up expandable is probably sensible: there is a reasonable path to 'just the text'

@cfr well, if you are sure that nobody will use \cyfres in an expandable context, you're fine with \NewDocumentCommand

@DavidCarlisle fancyref. But, yes, I know. But that's a weird case. My code is littered with standard cases. I want to know if that's wrong.

@JosephWright So I need \NewDocumentCommand \cyfres {} {} \DeclareExpandableDocumentCommand \cyfres {} { \l_cfr_cyfres_tl }? That would never have occurred to me.

@JosephWright that is a reasonable argument but it makes (l3 documented versions of) \def\foo{xxx}\foo very different from \def\foo#1{#1}\foo{xxx} and I'm not sure if that's a good thing in terms of being able to document what happens. I'd be tempted to force the user to use a newexpandable.... declaration. I think. Possibly.

@cfr As I've said, at present I'd use \newcommand precisely because this is an area we still haven't really explored

@DavidCarlisle We are back with the difference between a function with no arguments and using the content of a token list variable: it's not always clear

@cfr @JosephWright There should be \NewExpandableDocumentCommand, in my opinion. And the manual should emphasize that \New... or \Renew... should be used in document code as much as possible; the \Declare... variant should be in a small footnote.

@DavidCarlisle The earlier case, where a command can appear in a csname, shows you may need different handling anyway

@cfr there is no absolute notion of truth here, so "right" or "wrong" is not easy question to answer.

@JosephWright well if it had been declared as expandable it would have worked even with an argument, presumable?

@DavidCarlisle ooh then I can procrastinate

@cfr As I've said and as @DavidCarlisle's also pointed to, we are trying to define best practice but that needs people to test stuff ...

@PauloCereda we knew that anyway

@JosephWright well, as I said, one could also really just enable public handlers of this. I would personally consider preferable \cs_new:Npn \cyfres { \tl_use:N \l_cfr_cyfres_tl }

@DavidCarlisle Yes, that's true, but variables covers more than tl data and then life gets really 'fun'

@JosephWright true

@egreg I've raised something like this on the team list: my suggestion was slightly different (use a starred form to mean expandable, which automatically gives us \New../\Renew..), but you may well be right

@DavidCarlisle oh no

Ni

@DavidCarlisle Feel free to reply on the team list to my mail :)

@JosephWright people do whole phd's on eta-reduction perhaps @PauloCereda is doing exactly that just now:-)

@JosephWright I'm not favouring using stars for non-obvious things. Even \newcommand* is confusing, in a sense

@JosephWright I was waiting until I had decided what my answer should be...

@DavidCarlisle Am I too geek for giggling after reading your witty lambda calculus comment? :)

@PauloCereda yes

@DavidCarlisle I actually applied a beta-reduction. :)

@DavidCarlisle Right_{this context} := L3 team will not wittingly render it broken.

@PauloCereda you could also apply beta-radiation, it would solve all your problems

@yo' Then this should be in xparse. I thought that the xparse stuff was meant to be used to create a document-level interface whenever possible. If that's not the case, it should be made a lot clearer. To me, this muddles two different levels which I thought I was meant to think of separately.

@yo' ooh I might need sunglasses

@cfr exactly true

@yo' Yes, could be

@egreg Not that I know anything about it, but from an end-user perspective, this would make sense.

@cfr Yes, agreed

@cfr I'll make sure the team get a summary here so this goes into the discussion: this is all useful and @DavidCarlisle and I are trying to help!

@DavidCarlisle Erm, remember my degree is in chemistry not in computer science: eta-reduction sounds to me like adding hydrogen to a double at the some random point along a carbon chain ....

Or perhaps lowering the number of bonds to a metal centre from a multi-valent ligand ...

@JosephWright Sorry. I do appreciate that. It is just that half of what's said means nothing to me and it is all stated as if it is the most obvious thing in the world. I'm not trying to be difficult. It is just that I thought I was doing things the right way and now I'm alarmed that everything is going to break. (I don't mean the fancyref stuff. I'd not even heard of expl3 when I wrote that code.)

@JosephWright I thought he meant your present would be arriving sooner.

@cfr :)

@cfr Quite understand, and mainly hoping you see that David and I are also not trying to be difficult :)

@cfr We probably need some formal position on this so we can cover it in the xparse docs

@JosephWright don't feel bad for not getting the remark, it's very nerdy. :)

@JosephWright I do.

@cfr they want to save energy. It's what I do by not writing my thesis. :)

@JosephWright I suppose the question is if expandable declarations should be the default at the document level, even for commands with arguments, not sure how big the overhead for that would be (or if you could always use expandable processing inside csnames, or pdf bookmarks for that matter)

@JosephWright That would be nice. Then I will change everything if I need to. (\cyfres may need to be expandable. Most of the macros I have of this kind don't, though. However, they probably could be expandable.)

@JosephWright I would also likely to know about the non-expandable case. Am I not meant to use \NewDocumentCommand there?

@JosephWright I know. it was, as @PauloCereda indicated, a technical reference not intended to be understood by chemists:-)

@PauloCereda Doesn't that only work if you do nothing instead?

@cfr that too. :)

@cfr in general commands declared using \NewDocumentCommand are not expandable, so you are not meant to do that. Previously they were expandable if there were no arguments, but it isn't clear if that is how it should be (and we broke it and should revert that change) or if you were just lucky previously. Or possibly we could decide that you were lucky but now we should keep it working anyway for compatibility reasons. Any of those outcomes is possible.

@DavidCarlisle But what about cases where I don't need expandability but don't have any arguments?

@egreg @JosephWright @DavidCarlisle Why not a command \NewDocumentVariable without argument?

@cfr \NewDocumentCommand is fine, but the problem is that in general while you may know you dont need expandability it's impossible to document to the user when that is needed, especially considering cases like your fancyhdr one where the variable is passed to external code. It becomes like \protect in 2e where it is easy to use but the only documentation you can give, is "if something bad happens try \protect " so it might be better if we try harder to make more things "just work"

@cfr that's fine. I define \ProcessBlaBla this way

@DavidCarlisle No, they were not expandable, they were defined such that inside a \csname they would work if the replacement code was expandable: \csname ignores \protected

@DavidCarlisle Of course, as we have \unexpanded for writing files one might argue that \protected status is not so important if we can get to a new format, but I do note that ConTeXt have taken a similar line to xparse: \protected by default

@DavidCarlisle Cf. my question about contexts

@JosephWright yes that's what I meant ("expandable" is a bit of a vague term:-)

@DavidCarlisle s/impossible/requires significant thought/

@JosephWright yes that's what made me think of bookmarks

@DavidCarlisle Sure, though there we have to know that's what's happening

@JosephWright yes although I suspect that if it takes "significant thought" to come up with documentation as to when to use which form of declaration, most users (who don't read the doc anyway) will get it wrong.

@cfr, @yo', @UlrikeFischer I've summarised for the team: hopefully we can take this into account in a release soon

@DavidCarlisle Probably: I guess by 'user' I mean 'package author' and there we might hope to get people who learn expl3 to follow some guidelines

@JosephWright Possibly \NewDocumentVariable as an alias to \newcommand without parameters; or \NewDocumentShorthand (choose an appropriate name). The mixture between “functions” and “variables” should be avoided. Just a few examples can explain the differences.

@egreg I have uploaded a new package version this morning, to remove the \global references to \setlength. Thanks for the tip.

@StevenB.Segletes One less from the list! :)

@JosephWright @DavidCarlisle Thanks.

@egreg you can tell Frank about theorem.sty :-)

@PauloCereda -- no, no more procrastination. ignore @DavidCarlisle's comment. it doesn't apply to your thesis. it's bad advice.

@barbarabeeton aye aye cap'n! How's Gordon? :)

@JosephWright thanks!

@PauloCereda -- gordon is much better, but still shouldn't indulge in strenuous exercise, like shoveling snow. (we got another 10-12cm by late yesterday afternoon. then it turned to freezing rain, and now we have a half inch (about 1cm) of ice. i think we need to get a flame thrower.)

@barbarabeeton ooh

@barbarabeeton flame thrower ... or grit

Reminds me: Couple years ago, a water pipe broke atop a fill in freezing weather. The whole important freeway--tramway crossing got couple cm of ice on it, and also a tram line down the hill.

@yo' we could organize a big BBQ

ooh food

@PauloCereda had a nice goulash (actually more a perkölt) with dumplings for lunch

@yo' I had kabocha. :)

@PauloCereda Google, help!

@yo' <3 It's a Japanese winter squash. :)

@PauloCereda It looks like a pumpkin. Therefore, in my point of view, it's a pumpkin.

@yo' I am no specialist in pumpkins. :) I just eat.

@wilx well, IMHO much more sensible is \mathrm{i}, which is also what I use.

@yo' -- our next-door neighbor has ice rims of 2-3cm at the front of the steps up the small hill to his house. this morning, he was trying vainly to chip those away -- somehow the tread of the steps was relatively clear. since there's no handrail, and the steps are a bit steep, grit isn't enough.

@PauloCereda I have some tex code for pumpkins, you should study it

@DavidCarlisle ooh more procrastination

@egreg I have included this in my summary for the team

@Mico I am using apalike

@yo' -- the weather here has been so bad that on three separate days the public transit system canceled selected routes, and then shut down entirely until the streets were cleared. one of the routes cut was to the main hospital, where the entrance is at the top of a narrow winding road up a moderately steep hill. bad luck if you needed to get there by bus. (the emergency entrance is different, and easier for ambulances to reach. but not served by public transport. all buses here.)

@LuísdeSousa Mico will not see that @ ping as he's not been in chat recently, if this is following a question you need to do the ping there. (as a general rule, if the system does not pop up a name after @M then that name will not work.)

@DavidCarlisle Ok, I'll follow this on his answer.

So, has everyone voted in the election yet?

@JosephWright I am not everyone, but yes!

@JosephWright not yet but Karl told me my user name:-)

@JosephWright Wait, which election? TUG board or the meta question about TUG representatives?

@JosephWright How do I vote?

@DavidCarlisle :)

@ChristianHupfer :)

@ChristianHupfer TUG board

@PauloCereda Did you read the link I posted?

@JosephWright Then no...

@ChristianHupfer No panic though: you have until April

@JosephWright no...

@JosephWright I change to not yet then ;-)

@JosephWright: voted

@cfr For the present, Bruno's adjusted the code so your existing approach will work: I'll look to release later today

@JosephWright Not yet. Difficult decision.

@UlrikeFischer I know: I simply decided I had to give myself a 'slot' and reach some decisions in a limited time

@JosephWright ^^ :)

@JosephWright: you might enjoy this too: youtube.com/watch?v=yif3FM6PQtw

Snooker is really pure joy

@Martin I cannot play...

@PauloCereda I can not play either, but I really enjoy watching it

Hello guys

Would you help with references in Latex.

I don't know how to create a .bib file

@Cardinal 'ello!

Internet is full with the explanation on how to cite, not how to create.

@Cardinal Take a look at JabRef, it's an application.

@Cardinal it is plain text you just create it in any editor (although some editors have bib modes to help)

@DavidCarlisle not emacs

@DavidCarlisle Thanks, you mean I use the windows notepad ?

@DavidCarlisle Where should I save the notepad file?

@Cardinal well that would be my last choice of editor:-) but yes you could. What editor do you use for your latex file?

@DavidCarlisle not emacs

@DavidCarlisle Hmm, I am novice. I think you are talking about that IDE-thing, Texwork.

@PauloCereda go away and don't confuse the newcomers and don't mention vim

@Cardinal Ok I don't use texworks but plenty do, you could use that

@DavidCarlisle <3

@DavidCarlisle That's great. So, where should I put the .bib file?

@Cardinal simplest is just same folder as your document

@DavidCarlisle Shouldn't I do anything elese? forexample, choosing that file somewhere in the texworks.

I just did that and put the file around the rest of files.

I mean the files that are created by the Tex for each document.

lets consider this example.

Should I save the notepad file as "DUMMY.bib"?

you just want one bib file for all your references (to start with just for this document, some people have bib files with thousands of references for any reference they may ever need) then in your tex document you use

\bibliographystyle{plain}
\bibliography{yourbibfile}

then run latex, then run bibtex then run latex again (texworks almost certainly has a key or menu option to do that sequence)

@DavidCarlisle Oh, I should run bibtext separately?

I didn't do that. I thought the textwork does that in one session.

@Cardinal yes it is a separate program that is the point, the first latex run writes to teh aux file the citations needed for this document, then bibtex takes that list and fetches just those references from teh bib files, sorts them and writes a latex bibliography then that is input when you run latex again

@Cardinal it may well do, texworks may be set up to run latex and bibtex on one keystroke, you would be better to ask someone who has seen texworks about the details of that editor:-)

Ok, Fine. I was trying to understand the idea.

@DavidCarlisle: can I talk now?

@PauloCereda as long as it's not about vim

@DavidCarlisle no. :)

@PauloCereda I'm always happy to talk about emacs

@DavidCarlisle oh no

@DavidCarlisle It works perfectly.

@Cardinal happens sometimes:-)

:-)

@JosephWright Thanks. I guess you'll let us know when a more permanent policy is agreed? That is, would you mind?

@cfr I'll post something once we have this sorted: I suspect we'll have a documented idea soon-ish

@SašoŽivanović AS per meta.tex.stackexchange.com/questions/7212/… are you happy to go into the 'election'?

@JosephWright Sure, many thanks!

@SašoŽivanović Cool: I've leave the answer 'live' then

@ChristianHupfer Exaggerated! The number of thesis document classes is just short of \maxdimen, that is, 2147483647

@egreg I like to exaggerate ;-)

@DavidCarlisle I like r1377 ;)

@JosephWright :-)

@DavidCarlisle You must have spent a age working that out!

@DavidCarlisle I'm sure we all needed to know what happened in release 1994/06/01

@JosephWright easy enough given an editor that can read files form gzipped archives:-)

@DavidCarlisle I assumed you'd got the CVS or something to use

@JosephWright more fun reminding myself of that than working out how to make an asterisk in times roman.

@DavidCarlisle BTW, can you access the 'non-public' SVN on comedy? I assume it's got some of that

@DavidCarlisle :)

@JosephWright no, just looked in the base.tgz files from tug

@JosephWright in principle yes (although I wouldn't be too confident that we tagged releases so not always easy to see release versions from the cvs/svn log)

@DavidCarlisle True: we still don't ;)

@JosephWright seems like we are about ready for a PL2 though?

@DavidCarlisle Sure

I like such users

@JosephWright tex.stackexchange.com/posts/comments/869730?noredirect=1

@egreg Yes, I saw that

@JosephWright cultural heritage seems to be a theme this evening

Hi everyone!

Quick question, I'm looking for a command like \begin{verbatim} \end{verbatim} that's in-line

Something that, within a sentence or paragraph, ignores LaTeX and gives me that verbatim/lstlisting font for a single word

@Mathemanic \verb|\foo{}|?

Ah! perfect! That's just what I needed, thanks Egreg

Hmm. It's a little bit unusual that this verb command is written like \verb|...| instead of \verb{...}, isn't it?

@Mathemanic Peculiar behavior: you need a starting character that's matched at the end; it must be a character not appearing in the string to be typeset verbatim, not necessarily |.

Ah, interesting! So \verb{...{ would work but not \verb{...}

@Mathemanic no you can't use { but it has to be unusual as otherwise you would not be able to do \verb|}{| with unmatched braces