Ok, I just had a crazy idea; Text adventures are cool and can have really great descriptions, but generally you have to read them in terrible monospaced fonts, or terribly broken graphics. What if someone made a text adventure engine that used a version of LaTeX to render each paragraph as needed on the fly? You could use TikZ or pdfplots for the maps and diagrams.

Should I ask this as a separate question?

@DavidCarlisle I'm making a start on a source for the various LuaTeX things. Done so far just attributes and luafunction allocation.

@JosephWright OK I decided I needed to make some more examples (in particular to find out exactly what breaks if (as I suggested) luatexbase stops loading luatex.sty and just uses its own schemes. But turned out I had no time last night. If there are too many edge cases that change, and a "full" version is doable in reasonable time, might be best just to do it in one step after all. Elie seemed happy about the general plan.

@DavidCarlisle I'll see what I can pull together from the two packages, with some notes in a source, then I guess some testing will be needed with no-op versions of those packages.

@DavidCarlisle Some parts are clearer than others!

@JosephWright yes:-)

@DavidCarlisle Probably only need a few days to get a first draft, it will be the testing and so on that needs time

@FaheemMitha I would say yes, ask a follow-up.

@DavidCarlisle For example, I'm not clear at the moment why luatexbase uses Lua code for the attribute allocator. It can all be done from TeX and everything should I think still be Lua-accessible. That sort of thing will probably need explanation from Elie.

@JosephWright yes, (just like the day job:-) but I think I'm coming back to the conclusion that same testing is needed even for the simple change I suggested. So might not be worth doing. The change itself is fine I think, but answering the question about what exactly would change/break from the current setup is harder.

@DavidCarlisle Certainly

@JosephWright yes I thought that looked odd too. Of course some of that is old and it may be that luatex didn't used to have the convenience functions to access values of tex registers from lua?

@DavidCarlisle As I said, my plan at the moment is to get something down, then load that code with loading of the 'real' luatex/luatexbase blocked then see what happens with fontspec :-)

@DavidCarlisle Could well be: one question we'll need to address is what LuaTeX versions to cover (I'd favour if at all possible saying that 0.40 is a minimum)

@JosephWright which strangely enough is exactly what I did at the weekend with my version:-)

@DavidCarlisle :-)

@JosephWright 0.80

@DavidCarlisle Well yes, that might not be a bad position except what about people who do download the macro code outside of a TeX system update

@JosephWright we can probably set it up so that an explicit \usepackage{luatexbase}just over-writes any new stuff and does what it did for older luatex, just as an explicit \usepackage{etex} does what it always did.

@DavidCarlisle Yes, I'd hope so but luatex might be more tricky (Heiko uses \newcommand)

@DavidCarlisle One immediate question is whether we need to allow for both luatex and luatexbase syntaxes

@JosephWright no

@DavidCarlisle OK, that would make life a bit easier :-)

@JosephWright I would choose a syntax that you/we want, so informed by but not necessarily following either. Can always add on a compatibility package faking older syntaxes as need be, a bit like we dropped all the etex.sty localloc stuff but ealloc.sty puts most of it back again layered over the new system

@DavidCarlisle Fine

@DavidCarlisle I will though check over TL later on for the various usages

@JosephWright good plan

@DavidCarlisle OK, on that basis do we go with \newattribute (more natural) or \newluatexattribute (cf. \newXeTeXintercharclass and naming of primitives follow Karl's plan)

@yo' ok

@JosephWright don't much like having the engine name in the command as found with the mathchar extensions it's inconvenient once they are implemented in two engines:-) same as \pdf... fine for pdf related things like \pdfliteral but less so for pdfstrcmp. On the other hand we've gone with using luatex in the primitves and it's what luatexbase uses mostly, so probably I'd go with that

@DavidCarlisle As you might remember, I'd prefer to drop the entire luatex-in-names business too, but Karl was very unkeen at the primitive level

@DavidCarlisle Probably have to see what everyone thinks: for the moment I'll follow luatexbase

@JosephWright catches everyone out when they read the luatex manual and try stuff in latex...

@JosephWright yes think that is probably the thing to do initially at least.

@DavidCarlisle Yes, hence my position that we should really push to fix this by getting the small number of package name clashes fixed and then use the primitive names directly

@DavidCarlisle Roll on L3 :-)

@DavidCarlisle OK, I see why they did things from Lua: not necessary, just 'nice' (as they make a Lua table of attributes). My feeling is we can ignore that :-)

yes I can't see why you couldn't instead just have a convenience lua function that get the information back from tex...

scalar, vector, Carthesian and Hadamard products -- oh my goodness, this starts to be tough on notation.

@yo' just use juxtaposition for all products and assert that a competent reader will infer the correct operation from the context. Then it's not your problem.

@DavidCarlisle it will be my problem when they don't accept my thesis.

@yo' such practical considerations:-)

ah! I can assume that multiplication with no symbol inbetween is the Hadamard product! Yay!

@DavidCarlisle Progress is happening: over coffee I've started reading up on catcode tables :-)

@JosephWright oh no!

:)

@PauloCereda Next kernel update being planned

@JosephWright Already? :)

Hopefully @heikooberdiek will take an interest :-)

@PauloCereda It was already planned last year: we are now implementing the ideas

@JosephWright Oh cool.

@PauloCereda We are hoping to integrate LuaTeX-specific allocators and the like to remove the need for the luatex/luatexbase packages (at least the low-level parts)

@JosephWright maybe it is for support declaring attributes from Lua code? (I haven't looked at the code so maybe I am wrong)

@michal.h21 That is there but doesn't set the TeX part, so it seems a bit risky/odd. The main reason is for utility, which could be covered outside the kernel by a small package (perhaps team-maintained)

@JosephWright if I read this correctly:

\allocationnumber
\luatexbase@directlua{tex.write(
luatexbase.new_attribute("\luatexluaescapestring{\string#1}", true))}%
\ifnum\allocationnumber>\m@ne
\global\luatexattributedef#1=\allocationnumber
\wlog{\string#1=\string\luatexattribute\the\allocationnumber}%
\else

then new attribute is registered with luatexbase.new_attribute and the allocation number is then used in l\luatexattributedef. I think the call to luatexbase.new_attribute is needed because it saves the attribute name in Lua table so it is accessible from Lua code

@michal.h21 Yes, that's how they do it

@michal.h21 The same can be done from the TeX end and you can get the number out for use in Lua

@michal.h21 For the kernel the question is what makes sense as a 'minimum' provision

@JosephWright you need to get both number and name from Lua, I think it wasn't possible with vanilla Luatex's \luatexattributedef, this is why it is saved in Lua table in luatexbase

Hi everyone. I have a quick question regarding good practice when creating a new format using MiKTeX's GUI (Start →Programs→MiKTeX 2.9→Maintenance→Settings). Anyone willing to spare a minute? Should I perhaps post a new question?

@nnunes don't ask to ask, just go for it. If it's too involved for chat, someone will tell you and then you can post a question on the site.

@all do you know a dupe herefore: tex.stackexchange.com/q/250536

Or in other words, who do we call such macros, which do not delete spurious white spaces at the end of the line?

e.g. \label is not one of them, but \only seems to be one. I forgot the right phrasing for such behaviour

@LaRiFaRi it's not really a feature of the command, more of the space. spaces in horizontal mode make spaces..

@DavidCarlisle Sure, but there is some word for commands which do prevent that behaviour

If I knew some formulation, I could search for a dupe on this topic.

If not, I will answer and have one more code rewiev post here...

review

I could link to this tex.stackexchange.com/questions/7453/… but this is not really a duplicate as the user would not get an answer to her/his very problem

@LaRiFaRi "commands that have \ignorespaces in their definition" ?

@DavidCarlisle Ok, thank you. I will search for this.

@michal.h21 @JosephWright I don't think the table is really needed now the situation is completely analogous to count registers. If you use \luatexattributedef\qqq=300 from tex and want to access the attribute with name qqq from lua you can use tex.attribute.qqq which is 300

@LaRiFaRi yes that's a general issue, with that one (and some other "standard" answers) we tend to use that as the dup for any "% at eol questions, but really that's explaining the answer not duplicating the question

@DavidCarlisle I will ask a new question hoping that I will get an answer to which we can link.

@LaRiFaRi if the question is about % at end of line, we'd better leave it for @egreg to answer. It's only thing he knows about.

@DavidCarlisle I will just ask on how to replace each space in the MWE by some mark. Red rule or alike.

@DavidCarlisle If I get some non-lua-answer, this will be cool and easy for debugging stuff

@DavidCarlisle you will get value of the attribute, but not it's number. when you set \qqq=33 then print(tex.attribute.qqq ) will print 33

but you need to know the attribute number so you can test it in node callbacks with node.has_attribute(n, attr_number)

@michal.h21 hmm yes but... more testing...

@DavidCarlisle attributes are almost always used in node processing, at least I use them there and tex.attribute.qqq is unusable for that

@DavidCarlisle Here we are: tex.stackexchange.com/q/250551 I hope, it is understandable.

@egreg David says, this one is for you tex.stackexchange.com/q/250551 :-)

@michal.h21 You can get the number using \meaning, which can be put into the right place in some Lua code if required

@DavidCarlisle Exactly

@michal.h21 @DavidCarlisle's point about \count I think holds: if you want to provide access to the register numbers as a Lua table it should be done for all registers and therefore probably using callbacks for \countdef and so on

@michal.h21 One thing to bear in mind is that for the kernel we are really talking about what is required to make things workable. An allocator of some sort is required, setting it up to pass data back to Lua could then be added by a package without breaking anything else.

@JosephWright that seems OK to me

@michal.h21 Remember this is the early stage of this for us: first we'll try to map out some minimal ideas, then raise more widely. Will almost certainly come up on LaTeX-L once we have a plan!

@LaRiFaRi My answer here applies:

@egreg Thanks, this applies to my question on how to get rid of the whitespace. David already named it \ignorespaces... But it does not apply to my most recent question? Does it? I would like to get an easy debugger for users, who are surprised about "strange horizontal movements" and miss-alignments.

@JosephWright some possibility:

    \newcommand\saveattrnumber[1]{
      \setbox0=\hbox{\meaning#1}
      \directlua{
        texio.write("helllo world")
        local utfchar = unicode.utf8.char
        local name = "\luatexluaescapestring{#1}"
        % replace backslash
        name = name:gsub(string.char(92),'')
        local t = {}
        for n in node.traverse_id(37,tex.box[0].head) do
          table.insert(t,utfchar(n.char))
         end
         local s = table.concat(t)
         local number = tonumber(s:match("([0-9]+)"))

A user asked for help with a titlepage, should look like the followiing. Any TikZ wizard around that can take care of it?

@michal.h21 Funny way of getting the number :-)

@michal.h21 I would naturally start with

\attributedef\test=1 %
\def\getattrno#1{\expandafter\getattrnoaux\meaning#1\stop}
\edef\getattrnoaux{%
  \def\noexpand\getattrnoaux\string\attribute##1\noexpand\stop{##1}%
}
\getattrnoaux
\edef\demo{\getattrno\test}
\show\demo

then use \getattrno in the \directlua call (as it will expand to a number)

@JosephWright too many \expandafters and noexpands for my taste :D

@michal.h21 You could avoid that bit and do the parsing in Lua, but it was more the use of a \box rather than just \meaning I thought was a bit odd

@michal.h21 You could just stick \meaning#1 inside the \directlua as a string and it would work

@Johannes_B Is the monstruous background required?

@JosephWright you are right, it is little bit easier then:

@michal.h21 :-)

    \newcommand\saveattrnumber[1]{
      \setbox0=\hbox{\meaning#1}
      \directlua{
         local number = "\luatexluaescapestring{\meaning#1}"
         local name = "\luatexluaescapestring{#1}"
         % replace backslash
         name = name:gsub(string.char(92),'')
         number = tonumber(number:match("([0-9]+)"))
         print("attr number", name, number)
         global_attr_table[name] = number
       }
     }

@michal.h21 For max robustness I'd use a group to set the escape char, but the idea is simple enough :-) As I said, probably not needed in the kernel (or at least discussable)

what does the manual mean when it says tex table has "2 times 5 virtual sub-tables" and why can't I iterate through tex.attribute with for for k,v in pairs(tex.attribute) (@JosephWright, @michal.h21)

@DavidCarlisle I have no idea :D but attributes are probably userdate, like nodes itself and it can't be traversed using standard table functions

@michal.h21 I discovered your last statement is true:-)

@michal.h21 oh in fact that is documented (first section of 4.14 :(

@DavidCarlisle yes, I remember that it is somewhere in the manual

it seems that Simon Cozens is playing with constraint solving algorithms for layout in SILE:

github.com/simoncozens/cassowary.lua

it might be interesting to use it in output routine

@egreg Would be funny. A completely blurry titlepage in a bachelor thesis looking like a screenshot. :-)

@Johannes_B If the background is not required, a titlepage with \newgeometry or whatever is what's needed.

@egreg I always recommend to set up a new document and include it via pdfpages if different margins are in use. Easier that way (i think).

@Johannes_B Which is what frontespizio does.

@egreg Setting variables that are printed on the page, where/what is decided using options. Did i get this right?

@Johannes_B More or less.

@michal.h21 Probably you could jsut use the fact that the pattern is ?attribute\d+

What's the standard reason to close a question in which the OP found the problem while building a MWE and won't possibly help other users? This is the concrete question:

@GonzaloMedina Off-topic

@JosephWright Thanks. Could you close it using your powers?

People, do you recommend me a KOSS headphone?

Is it worth opening a question on meta in regard to questions like this one: tex.stackexchange.com/questions/250500/…

There are a couple of concerns and I think this is a good example. Say, in this case there's a question (albeit not a very clear one) which asks for a localized change. I think this is often referred to as the XY question. (Ask for X but need Y.)

I don't know how the format should deal with this but it seems strange that the current state of affairs also upvotes answers that are not really answers to the question itself, just workarounds that omit the necessity of answering that particular question.

Another concern to be raised is the way the question is asked: should the headline be improved yes or no, and how should it be improved for optimum efficiency in both search engines local to SE as well as 3rd party engines.

In this particular example "Display Math typeset as Inline Math" seems quite awkward, and actually kind of feels like the question is asked in reverse (the OP seems to ask for "inline math to be typeset as display math").

I don't know whether this has been addressed previously and whether it's something that could or should be brought up again.

Belated happy birthday @PauloCereda

@1010011010 The title was indeed misleading, I changed it

@1010011010 awww thank you. :)

@PauloCereda Are you in SP?

@egreg At home, today. :) I will go tomorrow. :)

@PauloCereda By shaking bus?

@egreg Always. :) I could bring some milk with me and return with butter. :)

Good news! Including PNG files with TL 2015 no more shows the annoying warning

libpng warning: iCCP: profile 'icc': 0h: PCS illuminant is not D50
libpng warning: iCCP: profile 'icc': 'RGB ': RGB color space not permitted on grayscale PNG

@egreg Yaaay!

@DavidCarlisle Oh my!

Hello guys

@DavidCarlisle How do you say this in english: "A prosthesis smart fitting solution" or "A smart prosthesis fitting solution" ?

@MarioS.E. the second (assuming it is the solution rather than the prosthesis that is smart)

@DavidCarlisle the prosthesis is not the smart one, is the solution that is smart

A smart-fitting prosthesis solution!

@PaulStiverson I like that one also hehehehehehe

@MarioS.E. but in that case it is the fitting that is smart. trouble is smart isn't an adjective I'd normally apply to a solution so it sounds a bit strained.

@DavidCarlisle I know, that's why I'm having so much trouble sorting this out. They really want to use the buzzword "smart" as in "smart solution for prosthesis fitting"

@MarioS.E. that sounds much better

qqq is attribute: 301 value: 53
hhhz is attribute: 444 value: 99
zzz is count: 200 value: 2

@JosephWright, @michal.h21 ^^^

@DavidCarlisle Code?

\luatexattributedef\attrzero=0

\countdef\zzz=200
\zzz=2
\def\zz{}
\luatexattributedef\qqq=301
\qqq=53
\def\hmm{lll}

\luatexattributedef\hhhz=444
\hhhz=99


\typeout{}
\typeout{}

\directlua{


v=tex.hashtokens()["qqq"][2] - tex.hashtokens()["attrzero"][2]
texio.write_nl("qqq is attribute: "  .. v .. " value: " .. tex.attribute["qqq"])


v=tex.hashtokens()["hhhz"][2] -tex.hashtokens()["attrzero"][2]
texio.write_nl("hhhz is attribute: "  .. v .. " value: " .. tex.attribute["hhhz"])



v=tex.hashtokens()["zzz"][2] -tex.hashtokens()["c@page"][2]

@JosephWright ^^^ :-)

@DavidCarlisle OK, so trivial with an up-to-date LuaTeX

@JosephWright perhaps although I don't see any documented guarantee that the internal command ids are really consecutive, but they seem to be...

@MarioS.E. -- "smart prosthesis-fitting solution" ties together the appropriate descriptors. hint: figure out what the most closely related pieces are, and tie those together with the hyphen. if there are a lot of descriptors (not in this case), it's sometimes appropriate to insert a comma between ones that are parallel.

@barbarabeeton Thanks! :)

@DavidCarlisle nice :)

@michal.h21 I might ask on luatex list if there is a documentable/stable version of that...

@DavidCarlisle Good plan

@JosephWright just getting rid of the latex remnants first;-)

@DavidCarlisle Good plan

@JosephWright I guess you saw the github conversation this afternoon?

@DavidCarlisle Yes

@DavidCarlisle Will circulate something in a bit to the team: have the mechanisms in place to handle registers, etc.: just the Lua side to think about

@Johannes_B I've closed per my comments and lack of feedback

@JosephWright luatex mail sent

@JosephWright Looking though the unanswered. Thanks for taking action.

@StefanKottwitz I had just come up with a solution with the Ishikawa diagram! :D

@JosephWright \ifnum\luatexversion<0.37 lose the 0. :-) but as I say I wouldn't support old luatex

@DavidCarlisle Oh, so it does :-)

@DavidCarlisle One for discussion, as I think I noted

Hi

@barznjy Still here?

@egreg no:-)

@egreg oh just spotted your comment, still since it's you I'll let my answer stand. Close as dup though?

@DavidCarlisle I voted so.

I need bug-hunting advice. Usually, if I am just patient enough, I can locate the source of problems in my own somewhat tangled code. Whether I can solve them is another matter, but I can at least find out roughly where the stuff causing trouble is. This time, I can't. With 2015, I'm getting American quotation marks rather than British. With 2014, no problem. csquotes appears to think I have invalid nesting even prior to \begin{document}. Commenting out different things helps with my MWE but

@cfr: Ciça is back! No cute kittens, but a very dangerous pyometra! Now she's safe.

breaks my document. Trying again, it'll work. Sometimes the MWE triggers csquotes to complain of nesting. Sometimes, I get complaints about unmatched braces. Commenting out various packages helps. And then it doesn't. And all that is in my MWE document body is <active quote>hi<active quote> where those are matched.

@PauloCereda Great...?! What is pyometra?

@cfr an uterine infection.

I tried using 2014's csquotes package but that didn't help. So it is not csquotes per se.

@PauloCereda Oh, dear. But she's OK? Just as well you heeded your vet's advice.

@cfr She's still drugged because of the cirurgy, but the vet said she is fine.

@PauloCereda That's really good. Your vet obviously gave you good advice.

@cfr add \tracingall, run in 2014 and 2015 and diff the logs, that's what I'd do

@DavidCarlisle Thanks. I'll try it. I suspect there is going to be way too much of it, though.... I can reproduce more-or-less without active quotes, but I do need csquotes, glossaries, forest and standalone, as well as a tangled mess I call a class.

As far as I know, TL 2015 used CTAN archives until some date at the beginning of May. I've uploaded some updates of two of my packages to CTAN a fortnight ago, so I did not expect those updates to be contained in TL 2015 right from the start, but I assumed they would be shifted to TeXLive soon, however, there are still the old versions in TL 2015.

@ChristianHupfer There will be a backlog of new stuff to deal with after the release, though. So you should expect it not to be as quick as if you'd uploaded at some other time of TL's life-cycle.

It also matters how 'easy' your packages are to deal with in the sense that some packages can be prepared for TL distribution using standard scripts, others have to be dealt with almost uniquely and then there is a bunch of other stuff in between those two extremes. At least, that is how it used to look.

@cfr Alright. I suspected it, but thought it would be quicker.

@cfr My packages are pure zip files, with a README, the .sty file, the documentation source, the manual pdf and the example pdf ... so I would say, it's quite easy ;-)

@DavidCarlisle I've got 4 log files. Two from putting \tracingall for the whole run - one 2014 and one 2015. And two from putting it right before the opening quotation mark in the document (so right before `hi'\n\\end{document}) - one for 2014 and one for 2015. However, I'm not sure what to look for. Definitely csquotes thinks nesting is open at the end of the document with TL 2015 and not 2014. But I don't know how to trace it back...?

@ChristianHupfer I just mentioned that as one factor. I suspect backlog is the main one :(.

@cfr Yes, I appreciate your answer. Is this backlog public, i.e viewable from internet?