@cfr "TeX is a language similar to TeX". Only a philosopher could come up with such a sentence... :) (I think you're being a bit hard, since the 'programmer' usage is much more likely an English error.)
An MWE makes your question more useful to other users with similar problems and almost always makes a question clearer. But what you are saying is that you don't have time to make a minimal contribution to what other people might gain from your questions, although you expect others to have the time to tease out of you the requisite details. OK. I'll add yours to my blacklist of questions I won't look at (nobody on it yet). And yes, you have every right to do as you do. Before anybody says otherwise, I do not deny that you have such a right. I merely assert my right not to read your questions. — cfr5 hours ago
@cfr: I wish I could give you 1000 rep for this comment
@TorbjørnT. It's not about the \clearpage being the answer. It's about the behaviour of the O.P. Asking questions, not willing to provide some assistance
@TorbjørnT. The problem is that not an obvious solution, there are key points missing. For instance he could have been using report and strangely set-up stuff about sectioning to force the lists to be sections. Or whatever else. And also, he's telling people to GTFO. He has the right to do so, and I have the right to add -1 to his question.
@UlrikeFischer Haven't quite got it clean yet but something like
@echo off
setlocal EnableDelayedExpansion
For /f "tokens=* delims= " %%I in ("%TEXMFHOME%") do (
set TEMP=%%I
set TEMP=!TEMP:%1=!
set TEXMFHOME=!TEMP!
)
endlocal & (set TEXMFHOME=%TEXMFHOME%)
will remove an entry from TEXMFHOME from the command line. (Need to sort out stray ; chars.) I'm not saying it's as convenient as for MiKTeX, but I think it is solvable :-)
Usage rmtexhome <path>
@UlrikeFischer Better:
@echo off
setlocal EnableDelayedExpansion
For /f "tokens=* delims=;" %%I in (";%TEXMFHOME%;") do (
set TEMP=%%I
set TEMP=!TEMP:%1=!
set TEXMFHOME=!TEMP!
)
set TEXMFHOME=!TEXMFHOME:~1,-1!
endlocal & (set TEXMFHOME=%TEXMFHOME%)
This is based around having separate entries
@UlrikeFischer Perhaps I should ask and answer a question on this (for the record)?
Then again, this is really Windows batch file work (and I guess I'd need to do the same for bash at least)
@JosephWright I know that it is solvable as I did solve it ;-). I have set TEXMFCNF so that is points to a special folder. And there I have a small texmf.cnf which is created on the fly and contains a TEXMFUF=all my pathes. And I have batch files which add and remove files from this list (and at the same time from miktex). But all this is not something for a newbie. You can sent a miktex user a texmf tree and tell him "store it somewhere, add it with miktex settings and it will work".
@UlrikeFischer I guess I'd be wary of what you say at the end there: adding anything to the local tree is an expert function, and I feel there should be exactly one local tree
@UlrikeFischer I'd rather send a TDS-style zip and say to add it to the local tree
@UlrikeFischer You could say much the same to a TL user, though: wasn't the issue not so much being able to add more than one tree and more the ease of removal of selected trees without having to reset everything?
@JosephWright "Thou shalt have no other local texmf trees"? Even in texlive there are at least two (home and local) and miktex users don't feel like this. And this started the discussion: I have a clear need for separated texmf trees. I don't want to confuse the datas of customers, I do want to be able to test a dev-version of biblatex without much fuss. And so I need both: easy adding and easy removing.
@UlrikeFischer I mean exactly one tree of each type: the TL-managed tree, the system wide local tree and the user tree
@UlrikeFischer You use case is relatively unusual, I guess: I just add and remove things from my local TeX tree as and when I need them
@UlrikeFischer I guess I'm mainly trying to say that I can see why TL is set up as it is: the need to swap trees in and out regularly is rare, and with the Unix background of TL it's expected that anyone needing to do that will be comfortable editing texmf.cnf
@UlrikeFischer @JosephWright it seems fairly rare (to me at least) you'd need a whole special tree with binaries and fonts etc) for one off jobs and testing packages etc I just put everything in a tree below the current directory and have .// in TEXINPUTS so don't really install anything anywhere, and all you need to do then to have different input paths for different projects is change directory.
@JosephWright Which means that they must now how to unzip something in a tree, how to find the (sensible) local tree. And in case that I'm only sending a test version they must know how to save and restore the previous version. You wouldn't believe how much time is can cost to explain all this small details so that they don't do something silly.
@UlrikeFischer Like I say, your situation is somewhat unusual I think (all of my code either goes to CTAN or would be expected to be in the same directory as a document)
@UlrikeFischer I do have some fun with achemso, where people do need local installs (managed systems with older TeX system versions)
@DavidCarlisle I'm not speaking about binaries. Most of my texmf trees are really small: Installation files for one font, for one class or package, or for one test version. And I have now quite some experience with such trees and I can say that they really pay of.
@UlrikeFischer I guess it's partly a workflow thing: like I say, I tend to just add stuff to my local tree then delete it again :-) Perhaps not efficient but I'm used to it
@ChristianHupfer ;). @TorbjørnT. Just to say that it may not be difficult to understand the solution, but I managed to misunderstand the question so it was certainly not obvious to me. But, really, I was reacting to the pattern of questions and responses rather than to this specific question. (& @yo' for completeness)
@cfr The questions of that guy tend to be unclear and there is not much help from him. It's sort of 'do it for me, you know exactly what I want to have'
@cfr Sure. As I (sort of) said earlier, I was looking at the isolated case, so I reacted a little, but if the general behaviour of the user is, ehm, less helpful, then that's another matter. Sorry, didn't mean to make a big issue of it.
@ChristianHupfer (ignore @egreg's interruptions) you can implement not either with etex \unless or simply swapping around the then and else clauses in the body
@TorbjørnT. No worries. I really just wanted to make clear that I had looked through the users' other questions before saying that, which I did not do lightly.
@JosephWright i was wondering if the bytecode allocator ought to start a bit higher (2 would do, but 10 might be more traditional, leaving the single digit ones unallocated) in particular this would allow you to byte compile ltluatex.lua (and include it into everyjob as a function rather than needing a file reference) if you documented the unallocated ones as kernel only...
@JosephWright Looks quite OK. But I'm not sure if roots should be separated by semikolons (in the texmf.cnf commas are used on windows). Also I wonder if you can add the same number of roots when using an environment variable then when using a texmf.cnf file.
@egreg @DavidCarlisle: Thanks for clearifying . ... We (some colleagues of mine and me) plan a LaTeX course for teachers and most of them are newbies, but I am expecting questions about \if... related content. I can't murder my participants with \int_compare:cTF ... well, I would rather be killed then ;-)
@DavidCarlisle: Swapping and \unless does the job well. I don't mind e-tex ;-) I think, for a reasonably new installation it's no problem to enable the e-tex enhancements
@JosephWright although actually I suppose it doesn't matter if you use bytecode register 1 to compile ltluatex and that gets overwritten by the allocator as by definition it would already have been used by then to run the ltluatex code.
Nach dem Stutzen des Rhabarberbarbarabarbarbarenbarts geht der Rhabarberbarbarabarbarbarenbartbarbier meist mit den Rhabarberbarbarabarbarbaren in die Rhabarberbarbarabarbarbarenbartbarbierbierbar zu Rhabarberbarbarabarbarbarenbartbarbierbierbarbärbel, um sie mit zur Rhabarberbarbarabar zu nehmen, um mit etwas Rhabarberbarbarabarbarbarenbartbarbierbier von Rhabarberbarbaras herrlichem Rhabarberkuchen zu essen.
@Johannes_B: I think you should answer that Pagenumber KOMA question. A mere comment isn't enough, in my point of view, since it's not outstanding and be overlooked very easily
@UlrikeFischer I went with what works: may do some more testing this evening
@UlrikeFischer No idea on support via the environment compared to texmf.cnf: certainly PATH can get pretty long without breaking so presumably this is OK
> The “colon” and “slash” mentioned here aren’t necessarily ‘:’ and ‘/’ on non-Unix systems. Kpathsea tries to adapt to other operating systems’ conventions.
Windows uses ; for separating paths so I *think it's safe. I'll try , too :-)
@PauloCereda When the old German tribes found Bananas on nowadays Germany ground they decided to stay there ... so indeed, the Banana is a very German fruit :-P
@Johannes_B: Thousand things you should have done before you die... Number One: Making bbq sausages by your own hand... please fill in the other 999 thingies then
@PauloCereda Doesn't mean much to me. I remember Dennis the Menace, though. Kid with a stripy top, if I remember correctly, and now-outlawed methods of disciplining your children.
I'm always frustrated when I see "kuřecí steak se sýrem a broskví" in a Czech menu. (It means "chicken steak with cheese and a peach"). It's not a bad dish, but it's certainly not a steak.
Is there any standard way how to enforce a minimal width of multlined? I have a very simple workaround using \mathrlap, but it's really ugly :-( Update: Ah there's a 2nd optional argument!
@ChristianHupfer I forgot to add that "Talián" is a funny word on its own: Czech language knows it in some dialects, in Slovak, it's the official word for an Italian -- the initial "i" simply disappeared.
Hunsrückisch oder auch Hunsrücker Platt ist ein deutscher Dialekt, der auf dem Hunsrück sowie im Gebiet von Santa Cruz do Sul, Brasilien gesprochen wird oder wurde.
== Verbreitung ==
Der Hunsrück besitzt klare geographische Grenzen, das Hunsrückische allerdings keineswegs. Es greift einerseits über den Hunsrück hinaus, hat aber auch andererseits eine gehörige innere Differenzierung, so dass es eigentlich kein einheitliches Hunsrückisch gibt. Die wenigen schriftlichen mundartlichen Zeugnisse machen klare Aussagen nicht einfacher. Seit den 1990er Jahren wird mit dem Mittelrheinischen Sprachatlas…
Could people take a look at this edit? It is marked as a clarification and has been approved. But it is not at all clear to me that it is what the OP meant at all.
The original doesn't mention Adobe at all. And why Acrobat rather than Reader? I am inclined to think the edit should be rolled back unless the OP confirms it really was the intended meaning.
@cfr Looks plausible to me, from the wording it could appear that the editor and the OP is the same person. Reader was formerly called Acrobat Reader, and is still part of the Acrobat "family".
Nothing new here: the patterns have been generated by patgen from a list of hyphenated provided by OUP in the 80s (possibly early 90s), and some of the recommended hyphenations have changed since then. It can certainly be updated, I’ve been contemplating it for some time now.
@JosephWright If we do so, I would strongly recommend getting the actual word lists with the primary and secondary breakpoints rather than just breakpoints - having different types of hyphenation is not currently supported in the TeX world (except for some experiments with ConTeXt, obviously), but it’s one of the areas where we can expect, and indeed foster, progress in the not-so-distant future. Hence if we’re going to get new data, I think we should really aim at looking forward.
@JosephWright It doesn’t really apply there: hyphen.tex is frozen for all eternity, but we’ve never made such a pledge for the British hyphenation patterns. And remarks about how they’re “outdated“ come up regularly now.
@JosephWright I would imagine so. It’s common topic when talking about hyphenation. Note the Germans have a working group that actually has word lists with primary and secondary hyphenation.
@JosephWright We have no particular policy: generally speaking we take in any modification that looks reasonable.
@JosephWright Obviously if you’re a publisher that has to reproduce typesetting exactly that won’t fly, but in this case you have to freeze the patterns anyway, I don’t think we can make any commitment towards supporting this use case in mainstream distributions anyway. Not enough workforce, and we need to be able to make some changes for end users.
@PauloCereda Like David says. It doesn’t necessarily have to be compounds, by the way; for any longish word there may be such a distinction. German and English have dictionaries that computes them.
@JosephWright So to sum up: I’m more than willing to do work in this area, but I’d like to plan it ahead a little bit to reflect the current (and foreseeable) situation in engines. Note also that when the British English hyphenation patterns were originally produced, OUP provided the word list but didn’t authorise distributing it, only the resulting patterns. I’d be interested to see if we can negociate a better arrangement today, in our world of open source and open access and open data.
@ArthurReutenauer Same thing as the latexrelease business, really: 'completely stable' actually has a tendency to mean 'causes some users real issues' :-)
@ArthurReutenauer In Portuguese, our rule is to keep the syllabic structure intact. Then we can hyphenate any Portuguese word, no matter if it's the first time we've seen it. :)
@ArthurReutenauer OK, so a reply could say that: UK-TUG (as in you and me) are willing to look at an update but it depends on data being available and on technical considerations
@JosephWright Do you remember that email half a decade ago when Knuth himself informed the TeX Live list (via proxy) that a bugfix in Metapost had caused all his working copy of The Art of Computing Programming to reflow? :-)
@JosephWright Yes, well the legal angle is certainly the crux of the matter here. We do depend on OUP’s good will to get started. Of course we have a very strong point today, saying that the TeX community has provided a service to the wider world, since the patterns are now used in, for example, OpenOffice too.
@PauloCereda I’d be really interested to see how that works in practice. In my experience this kind of general rule is just not enough – for any language. If you really want to do a good job at hyphenating stuff, you need really precise rules, and inevitably a list of exceptions. Otherwise you can have dictionaries list the breakpoints.
@JosephWright Sure, I’ll ask Dominik Wujastyk for his contact at OUP. He’s the one who did the real job originally, as I understand.
For example, supercalifragilisticexpialidocious is "translated" to Portuguese as supercalifragilisticexpialidoce. This is the corresponding hyphenation: su-per-ca-li-fra-gi-lis-tic-ex-pi-a-li-do-ce. :)
@PauloCereda So you mean syntactic analysis of texts? Yes, of course, but that’s still not the point. We’re talking about typography here: it is not identical with linguistics (though related, of course), and, more importantly, it has a lot of idiosyncrasies that people don’t realise until they look closely into it. Take ligatures, or kerning. These are things that most readers don’t even notice at any point in their life.
@ArthurReutenauer yep indeed. Also, the syllable-to-hyphenation transformation is not at all clear. Sometimes there can be two hyphenation points between two syllables, in some languages. Not in Czech though, for instance, but OTOH, the pronunciation of syllables isn't really reflected in the hyphenation points :-)
@PauloCereda My point is that it is the same with hyphenation. There are a lot of tricky details that people wouldn’t expect until they have to produce the patterns (using them as part of a typographical system doesn’t count). It usually turns out that what they learnt at school, or what they thought they knew, was a little bit simplified, or just factually incorrect (in some areas). I have no knowledge of the Portuguese patterns but this has proved to be true for every single language.
(@PauloCereda every single language that I’ve had to investigate in the past.)
@yo' Well, you’ve got the eternal problem between syllables and etymological. The two might be contradictory. Then you’ve got edge cases where the spelling is funny so that the written syllable doesn’t fit with the spoken syllable – and of course you should probably break according to the written syllable. Probably.
@PauloCereda Not necessarily. For languages that do have word lists computed, this is of course a very attractive option – that’s the case for German and English, for example – but for those that don’t, well the best choice may be to list the rules. There are not necessarily many, but you do need to list all of them, of course. For Turkish the rules are actually so simple that the patterns are generated with a script. So it really depends. You should look into our repository to get an idea :-)
@yo' Yes, but even then, the etymology may be in plain contradiction to the pronunciation. Take any word starting with “nad” or “pod” in Czech. Do you want to split after the ‘d’ to follow the etymology, or do you want to split before it if it belongs to the next syllable? And if you do the former you’ll have words starting with “nad” where the etymology is actually na + something else starting with d, so you’ll have to list it too (I wanted to say “nádraží“ but it doesn’t work).
@yo' Yes, it’s that simple. Well, someone needed to have the idea first, of course.
@ArthurReutenauer welcome to Math/TCS :) Many simple ideas give you a facepalm here :D
the example with .nad and .pod is interesting. There is no single word that would start with .poda and would have .pod as a lingustic prefix. It's always .po that is the prefix. Of course, if a consonant follows, it's clear that you don't break .pod
but then you have pod4nes because it's "po dnes" ("until today")
(btw, I'm just thinking whether there's an English word that would have 2 pronunciations and from that 2 different hyphenation points...)
@yo' Actually I meant things like .podl or .podr where .pod is the prefix. Here the syllabification would be po-dl[etc.], hence different from the etymology.
@yo' Yep! These three words (and, again, many others) can be both verbs and nouns. For some fascinating reason that I don’t know, the stress is on the second syllable when they’re verbs, and on the first one when they’re nouns; and that is reflected in the hyphenation: the verbs are hyphenated re-cord, pro-ject, pre-sent, and the nouns rec-ord, proj-ect, pres-ent, as if to put more graphical weight on the syllable that’s stressed.
2
@yo' I’ve never seen a source for the latter fact, but it’s the only explanation I can think of: the syllable that’s more important phonetically is made more important graphically by attaching more letters to it.
@yo' And then you’ve got groups of words such as “democrat”, “democracy”, “democratic” where the hyphenation looks completely random because the stress moves (and the breakpoints with it): it’s respectively on the first, second, and third syllables on these three words.
damn how can I use \showhyphens reasonably? I don't seem to get any hyphenation points for any words with diacritics, neither with pdflatex nor with xelatex :-(
@yo' That’s more for American hyphenation where they follow the pronunciation more closely; in British patterns they take the etymology more in account (but then again not completely either).
@JosephWright: Thanks ... it's about the same time as this year's event. Perhaps I get a leave for some day's earlier to quit and fly to Toronto then... (Looking at my bank account would help then ;-))
@JosephWright That's the only I reason I could not come to Darmstadt -- I got no leave from School board despite the fact there wasn't much to be done at school, just one week before summer break
@ArthurReutenauer Teacher of Physics and Mathematics, but a former Astrophysicist, but I've left university after my PhD, regarding \expandafter I am still a student ;-)
@cfr I do have one point to add: I think that nowadays the claim itself (that TeX documents are universally portable) is the problem, not the fact that it’s wrong :-)
@cfr The start date depends on the location. Since Germany is a federal state, each state has basically its own start time, but some states have in fact the some date, only Bavaria and Baden-Württemberg start always in the last week of July -- we're the last ones to celebrate the vacations ;-)
@cfr Or, to put it another way: it’s a very good thing to make people aware of the need for portability, a very bad one to make them expect it for free just because they use a system based on TeX :-)
@DavidCarlisle It would make more sense to me if TeX used the result of fontconfig's configuration rules but I don't know if that has much meaning on other systems. This would make more sense to me because it is, as I understand it, using my font configuration already. However, I would prefer it to do this only if the specified font is not available, rather than substituting it regardless.
@ArthurReutenauer I don't know what 'expect it for free' means here? Maybe a premium TeX to provide it :(.
@cfr Exactly what it says: to expect it without effort, just because they use a program with the string “tex” in it. This is, generally speaking, unrealistic. I have never seen it work when people actually needed it, and the few times when it would have worked it didn’t matter.
@ArthurReutenauer But it is one thing to say that you need to make sure you've got the same version of the engine, format and packages. It is another to say that you have to have the same fonts installed on your OS as well. If I want a document to do the same on, say, Mac OS X and Linux, right now I can pretty much do that. Once I use XeTeX/LuaTeX with system fonts, I can't do it without checking the contents of /usr/share/fonts and /System/Library/Fonts, /Libary/Fonts etc. ctd
But right now I at least get an error. If I substitute a font, I know it won't look the same. A list of fonts breaks that. My paper may be within the page limits when I compile it, but when you, the conference organiser, compile it, it may be rejected as exceeding the maximum allowed length. (And, yes, word counts are preferable. And, no, this is never going to happen to me because no conference I'll ever submit to nor journal will take .tex.) But there is a point there somewhere. ctd
@cfr not sure what you mean by "only if the specified font is not available" the test I suggested was testing exactly that wasn't it? using some built in font cache order might be more efficient but probably need to be difference for luatex/xetex in that case, this way you can abstract away the difference.
However, as long as nobody does this to pdfTeX and I can still use that, I am not terribly bothered as I only use XeTeX/LuaTeX for fun or family or friends and never for serious work. Though it would be nice to switch for when I need Welsh, but the results are manageable with pdfTeX even if not exactly ideal. (Until recently babel was essential although I see polyglossia has discovered the Celts.)
@DavidCarlisle Yes but I'd prefer it to rely on my OS's configuration if the specified font is unavailable. So no comma lists in the document. The document just says Times or whatever. If I don't have the font, then it uses fontconfig to get a match. But I wouldn't want it to use fontconfig to get a match initially since it would then substitute for Times. I'm not suggesting any of this is remotely practical. The 'I prefer' is a 'if I got to specify desiderata regardless of complications...'
@cfr My point all along has been that it’s extremely hard to actually have your document typeset 100% reproducibly between installations. The statement “TeX documents are completely portable” is just a myth, even with pdfTeX. Again, I have never seen it work without considerable effort. And your use case about submitting a paper to a conference is contrived: if your are the author and you’re aware of a feature that may make your document less portable, just don’t use it.
@DavidCarlisle Not sure I can bake it into the main commands (maddening complexity once you start considering ItalicFont=, BoldFont=,...) but perhaps a conditional? I'll try to come up with some use cases / requirements.
@cfr yes perhaps, but the version works there is just using classic tex tests so would (to make you happy) work just as well with pdftex:-) I don't see the exact details of the choice mechanism that important, the mechanism is mainly useful for documents that you want to be readable wherever not where you want complete typographic control, Like web pages, and the web has hardly failed as documents use different fonts on different systems.
@WillRobertson yes probably best to keep it separate:-)
@cfr That was also David’s. But going back to the proposed feature in fontspec, I get your (implied) point that it would be nice to be able to deactivate that feature entirely; thus editors of papers could use that option at the top of their master file.
@cfr About Polyglossia, I did notice comments by you on the main site saying that there were problems with Welsh. I was going to ask you at some point, but since we’re at it: can you tell me what they were? Or a pointer somewhere?
@ArthurReutenauer At the time, polyglossia did not know anything about Welsh. At least, it didn't seem to. I noticed recently that it now does but I have not tested it, so there may not be any problems. I don't know if it has inherited the problems with babel's support. If so, it shouldn't but it wouldn't be any worse than babel in that case. The main problem previously was simply non-existence.
@cfr It’s been there almost since the beginning, so I don’t know when you last looked. The GitHub copy only goes back to 2009, but it was already there at the first commit, and I suspect its existence goes all the way back to the first months of Polyglossia (I wasn’t following the list of languages closely then because I wasn’t in charge).
@ArthurReutenauer I just checked. The date specification is wrong in polyglossia, too. There is no ñ in Welsh. And the ordinal specification is altogether more complex.