@AlanMunn Wouldn't have thought so. I had to look it up. I've never discussed South Africa in Welsh and that's the only time I'd use 'township' in English, so this is not terribly surprising, I guess. Why? What made you think it is in Welsh English? It is still possible. My mother turned out to know a Welsh word I didn't because it was in Welsh English, even though she doesn't speak a word of Welsh. She can't say the word correctly, but she understood it.

@cfr I had no thoughts one way or the other, but I came across the word in /usr/share/dict/words. It's not in the OED, though so I thought I'd ask if it was possibly part of Welsh English. The word in English is common in the US (independent of S. Africa), referring to vague administrative units that are smaller than counties, but not quite full-fledged cities. The reason I found it is because it's an alphabetic triple (in this case containing the triple 'efg')

@AlanMunn I never heard trefgordd in the US. I hate to think how they pronounce it ;).

@JosephWright @AnyPowerThatBe Could you possibly unfreeze the Forest room again? I'd like to ask something and that's probably the tidiest place to ask it.

I wonder if it is possible to write some kind of script to say something in a room every 10 days or so. I keep forgetting to keep the room alive and I'm sure it is a pain to keep being asked to unfreeze it :(. Probably beyond my programming powers, though.

@StefanKottwitz I have a vague idea you are a PowerThatBe. If so and you're still around, could you possibly unfreeze the Forest room? Details ^^.

@AlanMunn The Academy dictionary says that trefgordd is for township in the historical sense and not for townships in South Africa. South African townships are treflan or maestref. Strange that tref doesn't mutate in maestref. If trefgordd is from tref and gordd, it seems to mean town mallet or town hammer. treflan looks like town up to me, but this is unlikely. maestref is something like town field.

@cfr The dictionary says the etymology is tref + cordd which means tribe, clan, family or troop. And 'trefgordd' is apparently the word used for the American sense.

@AlanMunn The dictionary didn't list cordd at all. The Academy one may, but I didn't know what to look up there. However, if so, it doesn't give it for tribe, clan, family or troop. I tried them all. However, I am obviously only using modern dictionaries. The Academy one lists obsolete words, too, but it isn't trying to cover older forms of the language.

@AlanMunn I've never used that dictionary before. Didn't know it existed. I like this:

> 14g. (17g.) AL ii. 692, Llyma fessur trefgordd cyfreithiawl: naw tei ac un aradyr, ac un odyn, ac un gordd, ac un gath, ac un ceilyawc, ac un tarw, ac un bugeil.

I don't understand it completely and the spelling is strange enough. (ss ??!!)

@cfr Google Translate says "Here is fessur township cyfreithiawl: nine and one aradyr tie, and one kiln, and one hammer, one cat and one ceilyawc, and one bull, and one bugeil." :)

@cfr The dictionary looks quite good. It aims to be the OED of Welsh. It's a shame you can't link back to specific entries.

@AlanMunn I think something like:

> Here is a legal measure of township: nine houses and one plough, and one kiln, and one mallet, and one cat, and one ??, and one bull, and one shepherd.

@AlanMunn Er ... roughly .... No real idea about ceilyawc - ceil y awc? Or ... But ...

@cfr I wonder if it's related to 'ceiliaf' which means to pen sheep together?

@AlanMunn That was my best guess, as well.

@AlanMunn You see the same root in bugeil (bugail is shepherd).

@AlanMunn Much too early/late. Please let me know if you puzzle it out.

@cfr Ok. Nos dawch. (Is that right?)

@AlanMunn Why are you interested in 3 consecutive letters anyway? Incidentally, it doesn't work here because ff follows f and precedes g in the dictionary sense of letter (as opposed to the linguist's, I guess). I'm not convinced by the un gordd -> one hammer. I suspect the gordd is cordd and is whatever the cordd is in tref + cordd.

@AlanMunn Yes. I'd say Nos da, but I've heard that version, too. Nos da.

@cfr Haha, caught by the Welsh letter thing again! :D It was a friend who posted a sequential list of English words, where each word had three consecutive letters, but they couldn't find a an 'efg' word. I did a grep through /usr/share/dict and found that word.

@cfr Yes, I agree, 'hammer' seems an unlikely thing. 'cordd' means family or tribe it seems.

@cfr Done

@StefanKottwitz no, just like LOT, LOF, or the bibliography, for some people, also the introduction.

@JosephWright Issue 26 is missing on the website: latex-project.org/news/latex2e-news

@UlrikeFischer yes @barbarabeeton and Karl made a few edits (looking for tugboat) and Frank will put it on the site today probably (it's harder than it ought to be to update that site unfortunately:-)

@egreg yes I was going to look what apacite was doing but I needed to pack up and drive here. quite a nice loop:-)

@cfr It is unfrozen

@yo' That cannot be compared. A list of tables doesn't list itself as an entry. In the toc it is welcome as it may need to be looked up (in contrast to the toc). A bibliography maybe doesn't list the book where it is in but can go to the toc.

@yo' But I take it easy. It's just talking about thoughts :-) we can see a toc in the toc, all equations numbered even without referencing (Count von Count of the Sesame Street likes that :-) ) and table cells with borders all around. Matter of taste or requirements. :-)

Hello everyone

Would anyone happen to know why the bibliography style parameter is ignored in pdflatex?

I've tried setting it to alpha to get those [REF69] citations

(instead of the minimal number in a box) though it has no effect on the output pdf

it's of course trivial, but I'd much prefer the references that hint at the name of the authors, I am referencing many papers but many of them are well known, and I'd like for the well known ones to be apparent immediately, without having to scan the bib.

I assume I've simply done something silly, I doubt I've found a bug lol

@SelflessSociopath I would guess because your input was incorrect. Best to make a complete small documented and post the full example as a question on the site. (Have you re-run bibtex after changing the style for example)

I actually never set up bibtex, I just do the bib. manually still lol

I tried including natbib package

@SelflessSociopath well that's why then of course. \bibliographystyle just writes the name of the style to the aux file as an instruction to the bib generator how to order and label the entries. If that is you then you have to do it.

ah, thank you!

@SelflessSociopath use bibtex, if you switch from numbers to alphabetic you need to add [abc2016] or whatever labels to every \bibitem and re-order the list into alphabetic order, which is a pain and error-prone, and a job that bibtex (or biblatex) were specifically written to do.

@yo' The "my teddy bear keeps interrupting my technical problems to tell me to murder more people" is probably a popular culture reference, but I'm not sure to what.

@DavidCarlisle lol I know it's a pain and error prone, I've organized 3 pages of references for this paper manually

I kindof was hoping to make them show up in chronological order, it's interesting to see these almost parallel universes of results and then the fields converge around '95

is there a convenient way to import all my hand written references?

@DavidCarlisle The advantage of not having to drive the train

@SelflessSociopath no, bite the bullet take the pain of making your reference a bib file then it's much easier after that.

@egreg but I got the tick anyway

Euh neuuuu!

Lol well thanks for ripping the bandaid off at least

:'(

@WillRobertson (and @DavidCarlisle, @JosephWright): Imho there is a problem with the Mapping-handling in the new fontspec version. The following document loads with TL15 the font

Arial/OT:script=latn;language=DFLT;mapping=tex-text;mapping=arabtex-trans-dmg;

and with TL16 the font

Arial/OT:script=latn;language=DFLT;mapping=arabtex-trans-dmg;mapping=tex-text;

The order of the mapping has changes and this seems to have the effect that the arabtex-mapping isn't seen anymore.

\documentclass[a4paper,12pt]{book}
\usepackage{fontspec}
\setmainfont{Arial}

\begin{document}
blbl \addfontfeature{Mapping=arabtex-trans-dmg} blblbl
\end{document}

@SelflessSociopath -- i don't think bibtex has a way to order references entirely chronologically. maybe biblatex has; i don't know. however, there's another option that allows (in fact requires) you to do your own ordering, and that's amsrefs. it doesn't get much use because (unfortunately), it has been made not easy to convert for use by other publishers. but it does allow one to specify the manner of referencing. you might take a look -- texdoc amsrefs on a tex live system.

@UlrikeFischer out of my depth really:-) I run your example and get no error but what was supposed to happen that didn't? oh hang on let me see in the log....

.
.  This font family consists of the following NFSS series/shapes:
.
. - 'normal' (m/n) with NFSS spec.:
. <->"Arial/OT:script=latn;language=DFLT;mapping=arabtex-trans-dmg;mapping=tex-
text;"
. - 'small caps'  (m/sc) with NFSS spec.:
. <->"Arial/OT:script=latn;language=DFLT;mapping=arabtex-trans-dmg;mapping=tex-
text;+smcp;"
. - 'bold' (bx/n) with NFSS spec.:
. <->"Arial/B/OT:script=latn;language=DFLT;mapping=arabtex-trans-dmg;mapping=te
x-text;"
. - 'bold small caps'  (bx/sc) with NFSS spec.:

@UlrikeFischer I get ^^^ (using the fixed tuenc.def)

@DavidCarlisle Sorry I should have added something that shows a difference also in the output. Add e.g. al-Alim` after the \addfontfeature and then compare the output of TL15 and tl16:

@UlrikeFischer don't easily have tl15 anymore, OK I'll look at the output...

@UlrikeFischer ah I get the second, with A which I assume is wrong

@DavidCarlisle See above my pictures. The first is from TL15 is correct, the mapping=arabtex-trans-dmg has been applied. The second is wrong as it uses (only) mapping=tex-text. I found the following workaround

\documentclass[a4paper,12pt]{book}
\usepackage{fontspec}
\defaultfontfeatures[\rmfamily]{}
\setmainfont{Arial}[Mapping=tex-text]

\begin{document}
blbl \addfontfeature{Mapping=arabtex-trans-dmg} blblbl al-`Alim
\end{document}

@UlrikeFischer yes that works here. Still. Seems like it isn't the fault of the format so I can make a sigh of relief and just blame @WillRobertson

I don't suppose anyone knows of a reason why my bibliography would suddenly stop putting book titles in italics? I know this is a pretty broad question, but they were all italic yesterday, and I can't think of anything relevant that I've changed. (Just added text to the document in various places.)

@Telemachus xelatex and it uses \emph in the generated bibliography?

@DavidCarlisle I do use xelatex, yes. Did \emph change in the last few days?

@DavidCarlisle Yes, as xetex can use only one mapping at the same time and the last one in the list seems to win fontspec should make sure that either there is only one mapping instruction or that newer ones gets at the end.

@Telemachus I'm too polite to name names but @WillRobertson broke it:-)

@Telemachus That's a bug in fontspec. It should be corrected soon.

@Telemachus the update went to ctan yesterday but for now just putting \let\emph\textit should get your italics back

Wow. I should never leave this room. Everything I wonder about ends up happening in here, more or less.

Related: latexmk changed the behavior of -jobname and that broke everything to hell, right? I almost hurt myself on Sunday over that.

I need to stop updating things until I finish this document.

@Telemachus ^^

Upvoted and thanks, @DavidCarlisle

As a followup, what will happen to book titles with book titles in them, I wonder...

E.g. <<A commentary on the <<Republic>> of Plato>>

Be right back with the answer...

@Telemachus if you are using 2017/01/01 release also (and this time it was me not @WillRobertson that broke it) . ``hello'' . will typeset as that and not ligature to real quotes, similarly --- will not make an m-dash. a patch is already on ctan but not yet in texlive, but you need to make sure tuenc.def gets updated before your do a final run, sorry.

@Telemachus yes as I indicated in the answer making \emph be \textit loses the ability to switch in nested contexts, the github link has some low level fontspec code for the real fix, or the fixed fontspec will be distributed via texlive today or tomorrow...

@DavidCarlisle can he just pull the single file from somewhere and put into the document dir?

@yo' well it's ltoutenc.dtx so you need to run the ins....

@DavidCarlisle Noted re ``foo'' etc.

So far they seem ok here, but I'll keep that in mind. And, yeah, re nesting with emphasis/italics. Got it.

@Telemachus yesterday here we were discussing how urgent it was to release an updated latex format and Joseph said...

:-)

Yeah. It's stupid of me to run updates in my context. Habits hurt, sometimes.

(Not my thesis. A text that needs to go to copy center for students early next week. Hard deadline. Same problem.)

The latexmk thing really hurt since it broke the build. Nothing would compile. Easy fix in the end, but took me a few hours to see it.

@Telemachus normally it's fine but this week saw in some ways the biggest update ever for xelatex and lualatex as we made them default to Unicode rather than classic 7bit "OT1" tex encoding, mostly it's gone OK but there were a couple of issues, the 2017/01/01 Patch Level 1 release should be in texlive by tomorrow and sort out all reported problems.

@Telemachus Yes I realise that, but same issue as you say

Sounds good. Relatedly, that sounds like a very good update. Thanks to all for the work.

@DavidCarlisle Does this switch of encoding depend on font support or is it purely LaTeX internal thing?

@wilx not sure what you mean, but font support is as before(almost) except that xelatex and lualatex start off in TU encoding and lmr family instead of OT1 encoding and cmr family. To make that happen for lualatex, luaotfload.lua is now loaded in \everyjob

@AlanMunn An earlier word than llwyth, maybe, which is what I'd think of. (teulu for ordinary family.) The Academy dictionary seems to speak a different language, but I guess that's because it is focused primarily on current usage.

@JosephWright @StefanKottwitz Thank you!

@michal.h21 did you get the ufy thing to run, I'm usually on cygwin which complicates all these multiple library setups, but I'd really like to see harfbuzz running in lualatex:-)

@DavidCarlisle Will the next version change \textasteriskcentered to

\DeclareTextSymbol{\textasteriskcentered} \UnicodeEncodingName{"002A}

I just saw a question about a disappearing asterisk

I wonder, when preparing a new LaTeX class for a journal, how long back shall it work? I would really like to use expl3, but many of the features are "new"...

@yo' -- do you mean "how much of the backlist should it be able to process with no problems?" probably only a few years. but if it's a class file that's to be used by a lot of people, if it's really significantly different, i'd give it a new name, and not disable the old version. (willing to discuss this off-line.)

@barbarabeeton I would give it a new name, that's no doubt. The problem is, it's used by lots, and you know how some people are stubborn...

@yo' -- stubborn! that's an understatement! we do try to keep the ams classes fully compatible, and for internal use, we use snapshot to record exactly which version of all class files and packages were used in production. so that if the need arises to resuscitate something (say for republication in collected works) we are able to "recover" the environment. so the only reason for a really radical change (and probably a lot of headaches) would be if the visual style had to be modified.

@barbarabeeton that's partially the case here (style modification)

what you really need to do is provide a warning message that the new class isn't/may not be compatible with old document files.

off for dinner, will return

@yo' It naturally depends on what the journal want. But I always try to get the message through that nowadays it is much more difficult to write a class that can be used by texsystem of very different ages, that is impossible to write fallbacks for all resolved bugs anyway and that if they want all sort of goodies (as they always want ...) then modern methods are needed anyway.

@UlrikeFischer yeah, I'll obviously have to ask the chief editor

@UlrikeFischer I changed it to \DeclareTextSymbol{\textasteriskcentered}\UnicodeEncodingName{"2217}

@DavidCarlisle I got ufy running on Fedora 21, but Luaharfbuzz doesn't work here. It is early stage of development, it can typeset only plain text at the moment.

regarding Windows, Deepak posted detailed setup for Luaharfbuzz compilation on Windows, but I didn't succeed in this either. I don't really like C libraries compiling, or C libraries don't like me, because they fail with me quite often.

Anyway, my harfbuzz shaper more or less works on Linux with recent LuaTeX, but the code is somehow messy and I don't understand half the code anymore :(

@DavidCarlisle Ah ok. Works too ;-)

The biggest issue is that you need to push text for the whole paragraph to Harfbuzz, and it is quite difficult to insert various glues, whatsits and other stuff back at the right position

@michal.h21 yes I must admit despite watching your and Deepak's github I never really tried to build this, practically speaking it's only feasible for latex if it can be built and linked on all platforms, would be easier if it was statically linked like xetex

@michal.h21 whole para or just runs of consecutive characters (which is what xetex does as far as I can see)

@DavidCarlisle at the moment I shape only consecutive characters, but the issue is that in some languages glyphs may depend also on position in the sentence. So this solution is insufficient.

@michal.h21 hmm be interesting to try such a text in xetex

@DavidCarlisle it would be nice, but I think that LuaTeX team isn't in favor of this solution :(

@michal.h21 in the end that isn't completely their choice.

@DavidCarlisle I think Khaled could say more, but I am not sure if he want to have anything in common with TeX anymore

@DavidCarlisle they prefer SWIG for binary libraries, which is something I got to work, but can't say that it was easy

@michal.h21 true but he is very active in harfbuzz sources at the moment, but I just meant if you have such an example requiring sentence context can we just try it in xetex?

@DavidCarlisle no, I don't have, I was just told by Khaled that this is the main difficulity. I guess that there are some examples in Arabic script.

LaTeX News 26 latex-project.org/news/latex2e-news

$ git push
Counting objects: 10, done.
Delta compression using up to 8 threads.
Compressing objects: 100% (10/10), done.
Writing objects: 100% (10/10), 1.38 KiB | 0 bytes/s, done.
Total 10 (delta 7), reused 0 (delta 0)
remote: Resolving deltas: 100% (7/7), completed with 5 local objects.
To git@github.com:ho-tex/hyperref.git
   9143985..ddc3db5  master -> master

@JosephWright @WillRobertson hyperref manual uses lmvtt so i needed to make a tulmvttfd file....

@DavidCarlisle Goody

@DavidCarlisle Work PC now at least has TeX and Thunderbird: still a bit distracted by having to sort this!

@DavidCarlisle The latexbug package has been added at the right time! :P

@JosephWright Frank's reconfiguring mail as well, @egreg will have to send his bug reports to me:-)

@JosephWright I was just reading this work of fiction:-) anorien.csc.warwick.ac.uk/mirrors/CTAN/macros/latex/base/…