@PauloCereda Wow I didn't know about master and balance counters at all. There should be a couple dozen extra warning signs at the beginning of the "brace hacks" section...
@egreg: Great answer here: tex.stackexchange.com/questions/82554/… I wonder about a curious abomination.tex in your filesystem. It certainly has fncychap, I believe. :)
@DavidCarlisle Oh, I thought you were trying to be funny... So, should I post another question, as it seems you are saying that \relax can be bad if you are not in math mode, but I am trying to determine if I am in math mode so need to use \relax before \ifmmode?
@PeterGrill etex has a `if you can use to detect if you are in that omit pre-scan and so you can just add the \relax there which is safer (etex's a bit modern for me I'd have to look up the name)
@PeterGrill also \protected\def\foo{\ifmmode....} doesn't do a \relax in the unprotected case or perhaps I was wrong about the if, you can check you are in an alignment currentgrouptype=6but that isn't quite what you want
@DavidCarlisle Well it seems to be working for me (in the small test case, full document still recompiling), so will just add a note there for now to post a follow up question in case there are issues.
hmm I think if lastnodetype=-1 and currentgrouptype=6 you are almost certainly at the start of an alignment cell. Next challenge is to find somewhere else where that is true
You use File.separator because someday your program might run on a platform developed in a far-off land, a land of strange things and stranger people, where horses cry and cows operate all the elevators. In this land, people have traditionally used the ":" character as a file separator, and so du...
"where horses cry and cows operate all the elevators" LOL
@percusse At the moment there are a lot of German bands I like. It has become more popular to sing German (outside the "Schlager" genre) since the "Neue Deutsche Welle" :-)
@StephanLehmke As a Déformation professionnelle I follow the drummers rather than bands. I don't know if it is your thing but I enjoy listening to Sieges Even.
Ok, this has been bugging me for a while; Will cruft left over from a previous run of LaTeX ever change how my document looks?
Would the following procedures ever give me different results:
I compile my document until it stabilizes (All references resolved and such) then edit my document and ...
@percusse "Anything unnecessary or redundant; esp. poorly designed or unnecessarily complex computer software, e.g. that containing sections of obsolete code. Also: filth, esp. that which builds up over time; detritus." Used in the second version.
@percusse "1999 APC May 146/2 The cruft that accumulates in the front fan grille of a computer with no dust filter." or "2005 Wired Dec. 156/2 Tagger cross-references the acoustic fingerprint of each untitled tune... It was about 85 percent accurate with our playlist cruft."
@percusse For once I don't like the OED definition though, as cruft as I've heard it used is always in the form of that 2nd quote, as in, things that once were useful, but have built up over time.
@percusse That is; I'm not sure. It must be getting information from the .aux file, or compiling multiple times to get references and links right wouldn't work. So I was wondering if it would ever save old paragraph information or information about the placement of floats on a page, then not reset that each time if it didn't have to.
@Canageek It does consecutive passes over the same material but that's not going into the aux file. So better ask the TeX alrighties. :) No I've been introduced to English in high school via heavily smoking elder students and heavy metal people, we had also some courses...
@Canageek bad isn't quite the word I'd use. I'm sure I'd never right a bad package, but I'm reasonably sure I wrote longtable
@Canageek you get the same effect in anything that leaves measurements in the aux file (some uses of tikzmark for example would probably have this problem)
@Canageek Yes but they are not my thing to be honest. Actually I shouldn't say big words like that because recently I was going over Pain of Salvation's early albums and I've realized what a big mistake I did by dismissing them judging by their later albums.
@Canageek not really, unless you know the package really well, For instance if a package writes \def\foo{25cm} into the aux file that might be harmless or it might not. If for example it is recording the widest item (for some class of item) so that it can make some uniform heading that wide, then if you delete the wide item the package may insert the wide heading which forces all items to be that wode so they just get wider and wider. This was what would happen in longtable
@PeterGrill in the case of longtable it was a documented feature. The package refused to use the saved widths from previous runs by default unless you added \setlongtables and it was documented that if you edited the file without taking \setllongtables out bad things would happen.
@DavidCarlisle I have a feeling you purposely added all the issues in the table packages knowing that someday they will come in handy!! Can't believe you are already rep capped for the day!!
@Canageek well I'm sure that some uses of tikzmark (or tikz generally) suffer from this. That lets you measure the distance between two pints on the last run. They may be wrong if the file has been edited so it deoends how that information is used
@DavidCarlisle Yeah, but I got a table answer yesterday so was happy with that... I think I'll just ask a lot of questions, and if others answer then the extra 2 upon accepting that I get is not subject to rep cap so that is going to be my secret weapon... :-)
Of course if you answer I am down by 13, so have to avoid any table questions, or TeX questions, well,,, ok I 'll ask ONLY tikz questions. Haha found the secret... :-)
Is there any way of using the changebars package with XeLaTeX?
The current release of changebars doesn't support a xetex option. Almost three years ago there was a discussion on comp.text.tex about a patch to the package to support XeTeX, but nothing seems to have come of it; there was a simila...
I could use andrews tikzmark example but my way of getting to the margin from wherever the start/end marks were probably wasn't very tikz, and I couldn't see away of making it span multiple pages without using so much low level tex code I may as well do the rule by hand as well
@PeterGrill dunno he has an example coming with the package that does two pages but there is a specific tikz path call at either end so I couldn't see how to generalis thet to multiple pages without hacking the oitput routine to know that it was mid-change region(s)
@DavidCarlisle A very nice drive I hear. My brother did it once. I have a lot of family in that area, Mom is from Calgary, and my family is centred around Drumheller.
Anyone by any chance have a question where you need some line drawn from one place to another so I can give a \tikzmark answer, or now that David is sleeping, how about a simple table question??
@hpesoj626 New table question posted. If it does not get answered by the time @DavidCarlise gets up I'll have to delete it, can't make it too easy for him...
@hpesoj626 I don't think it is too localized (which I mostly disagree with). OP wants a multiplication symbol. It seems as if he does not realize how quickly one gets a response here, so hasn't seen the comments yet.
@PeterGrill Your problem is strange. Although if you try to put \textcolor{red}{Curie} in your .csv file, it works. I don't know why \IfStrEq doesn't work
@percusse Ah, so that is progressive metal. Sounds nice. With metal I somehow associate the sound of chainsaws ;-) Do you also like more "jazzy" stuff? (see 1:35:00 or 00:36:00)
@StephanLehmke This writes :locref "\textsc {i}" but this string needs to be read in by xindy, and from xindy's point of view that string starts with "\t" (tab character), so I need to escape the backslash.
Which is done at the start of glossaries \@@do@wrglossary but that's too soon if the page counter is in a page spanning paragraph
BTW I think it really needs to be done in TeX rather than post-processing the file. I get enough users grumbling about using makeglossaries as it is.
@StephanLehmke It needs to write :locref "\\textsc {i}" rather than :locref "\textsc {i}". All control sequences (except for the one that generates the actual number, such as \roman{page}) need to be \stringified and have the initial backslash doubled.
@hpesoj626 No, I hadn't seen it. Thanks for pointing it out.
BTW any solution needs to work regardless of what counter is used. For example, replacing \thepage with \theequation or \thesection
@tohecz triplet is (almost) exclusively used these days for one of three babies born at same time (like twin). triple is more common for the ordered list, along with quadruple for 4, ... hence n-tuple for a length n list.
@NicolaTalbot This may be a really silly idea, but what if you wrote out '\textsc{i}' instead? I.e., single-quoted and thus (possibly) not subject to expansion.
@StephanLehmke The \write0 was just for the minimal example. It's actually a \protected@write, and I don't have any control over what goes in there. A user may have a really wacky page number design.
@NicolaTalbot But it will come out of some glossary macro, won't it? Can the user write down the whole \protected@write construct themselves? How do you hook in to escape backslashes?
My first impulse is \protected@edef, then \detokenize, then escape Backslashes as characters.
@StephanLehmke Commands like \gls ultimately use \@@do@wrglossary. The counter used for that entry's location is stored in \theglsentrycounter (which defaults to \thepage). This gets expanded and stored in \@glslocref via \protected@edef\@glslocref{\theglsentrycounter}. If xindy is in use this then gets processed by ` \@gls@escbsdq`, which does the double-backslashing.
The samplexdy.tex file that comes with glossaries illustrates the problem. It used to work until v3.02 when I modified the code to fix the problem of the page number being out in page spanning paragraphs.
@NicolaTalbot I have some problams understanding that. The whole point of using \write is to make sure the page number is expanded at the right time. When is \protected@edef called?
@StephanLehmke To do the double-backslashing. I'd forgotten the problem about expanding the page number at the right time when I wrote that bit of code. What I was hoping to do was delay that bit of code until the \write, but if @Andrew's suggestion works that would be the best solution.
I recently posted a question Display issues drawing DNA sequences with TikZ. The issues I described were for the most part resolved, but one remains. I managed to simplify my example down to show the issue more clearly. Consider the following code.
In this code, the line drawn between the two mu...
@NicolaTalbot It would probably? be simpler if you never bothered xindy with the formatting of the page number it doesn't really need to know it does it? If it always just wrote out something like /glossarypageformat{5} with a arabic page number then xindy/makeindex could do sorting stuff and when you read it back in you could flip / to being the escape char and format the result, or something....
@NicolaTalbot protectedwrite has a spare arrgument for you to insert extra definitions so you could insert \let\theglossarycounter\realx if you wnat it to be held like page
@NicolaTalbot you could but you don't want to, changing catcodes mid document is evil and doesn't work (like verb) inside other macros, but you can use a character like / (or j if you like xii.tex) and then switch that to being escapechar on reading it back as you are in control there.
@NicolaTalbot It works best if you can spell relax though
@PauloCereda It's still holding glossaries to ransom, but @David had a good idea. But first I have to typeset some Bayesian inference notes before I can mount my trusty stead and charge off to the rescue.
@PauloCereda basically it's a typical Knuth-ian optimisation where if it is possible to use some bizarre combination of commands to achieve an effect then that is documented rather than adding a simple extra primitive. Alignments need a special kind of grouping so that when you go \begin{array} inside \begin{array} the inner & are taken as not ending the outer cell even before \begin{array} has been fully expanded and TeX knows that it is an \halign.
That particular combination basically acts like { and there is another one at the end that acts like } but it interacts with the pre-scan for \omit in a way that works rather than a way that doesn't work.
@PauloCereda The trick is used basically to insert "open" \noalign bits in definitions. For instance \hline does \noalign{\ifnum0=}...\futurelet\reserved@a\@xhline` to look ahead for another \hline; then \@xhline ends with \ifnum0={\fi}` to close the \noalign. In this way the brace counters that TeX uses in alignments is kept happy.
@PauloCereda Years ago (last century) a film crew came round to interview me about me work (using Bayesian belief networks to assess the risk of food-borne botulism). I'm sure the interviewer's eyes glazed over when I tried to explain Bayesian statistics.
@PauloCereda When the camera actually started rolling, all he asked was is there a button you can press that will tell you the food is safe? No. Why not? It just gives you probability distributions. But can it tell you there's no risk? No, there's no such thing as zero risk. Ok, let's just film you doing something interesting on the computer.
@NicolaTalbot There is something to be said for protecting \thepage that way. If the page number format is redefined between pages, then if \thepage goes into \write unexpanded, the page number will come out with the correct formatting.
@NicolaTalbot So I suggest following David's suggestion, but be careful what is expanded when. If the macro name you write out is your own, then you can just define it to expand to the escaped version and don't need any preprocessing at all.
@NicolaTalbot @PauloCereda was referring to this where it is all explained (the following one explains rocket science, and today's explains logic) xkcd.com/1132
@DavidCarlisle I think that sums up frequentists vs Bayesian nicely ;-)
@PauloCereda Mind you, it wasn't as bad as the time another film crew came round to ask about muesli, which isn't my field of expertise, so I just looked like a token female with nothing to say. I think I got cut out of the shot in the end.
@PauloCereda One needs to move in the right circles. You have an IPad like Andrew, I have a samsung android tablet like the queen bbc.co.uk/news/technology-20314882
@JosephWright There was a spate of film crews hanging around NRP in the late 1990s. In addition to the ones filming me at IFR there was also some filming going on at UEA for some TV series. (Can't remember which one now.) I remember them lolling very untidily on the footpath I wanted to use.
user19161
@NicolaTalbot Maybe the soup is too delicious so you did not notice when some of it splashed onto your glasses. =)
@WillHunting :-) Although I'm not convinced about that. There wasn't a great deal of choice in the cupboard. I've ordered some groceries online, but the delivery isn't due until later. (The nearest shop is 1.5 miles away.)
@JosephWright moderator question do we just close legal questions on sight rather than answering them?
The answer to tex.stackexchange.com/questions/82713/… clearly should just be a simple "no" in a sane world but I'm not sure about the world we actually live in.
@DavidCarlisle I believe that there is at least one question about that; but it's like having to license as GPL a program just because it was compiled with gcc (of course using some libraries might require such a license) or the output of a Perl script.
@DavidCarlisle In general, my approach is that these are (sadly) off-topic, as we don't have the necessary legal expertise to answer. The intention is I am sure as @egreg says, but I would not be sure on interpretation
@PauloCereda :-) I like your message. It's far more upbeat and cheerful than mine!
@PauloCereda I was also thinking about making a command line batch version, but I haven't used it often enough to provoke me into making the necessary changes. (I only use it just before uploading the PDF to the printers, so it's not like I use it every time I latex a document.)
@egreg Looks like the alternative approach to \prg_replicate:nn discussed in the LaTeX3 docs: I think your are probably right about this being David K.'s idea
@DavidCarlisle Well I guess something like this is used when you need something expandable in TeX, not otherwise. I haven't thought about possible alternatives though.
@StephanLehmke One reason is probably that it does the replication from the wrong end so for large input it is putting a lot of tokens into the input list which you then have to finally process in the right order at the end
@DavidCarlisle OTOH it only reads the string to be replicated a linear number of times. I'd expect this to be more difficult if you want to avoid that effect.
But I'd like to see how to "jump over" the input if you want to process the other way round.
Maybe if you're allowed to use \unexpanded, but then it'd practically only work in \edef.
@DavidCarlisle I had assumed the specification was "\replicate{n}{X} expands to n copies of X", which doesn't seem to be met by this solution because the first copy of X would be expanded/executed before the replication was over.
@DavidCarlisle Interestingly, your solution only seems to be slightly slower than the abdomination one. I'd have assumed otherwise, because it seems to read stuff multiple times.
@StephanLehmke ah but when do you really want to just replicate the tokens without expanding any of them:-) It does have the advantage of working rather than not working for big numbers though
@DavidCarlisle Certainly. No offense meant. There are surely reasons why the other one is called "abdomination". I just want to understand better. The recursion limit was the first disadvantage which occurred to me. Now I want to understand the others.
One thing that occurred to me was that the "abdomination" variant really expands away completely, leaving n copies of its argument, before the first copy is expanded. As I'm often faced with problems of "expanding over" some string (that is, expanding stuff behind a token sequence without touching the token sequence itself), I'm intrigued what other techniques exist to achieve this effect.
@StephanLehmke You'd have to try harder than that if you did mean to cause offence.:-) Also it is hardly a considered bit of code I just wrote it in the last few minutes so it's surprising it works at all:-)
@JosephWright Great stuff! But a bit too deep for me to understand just by looking at it (so I have to analyse). At first glance it looks very similar to the abdomination, only with logarithmic recursion depth...
Ok, this has been bugging me for a while; Will cruft left over from a previous run of LaTeX ever change how my document looks?
Would the following procedures ever give me different results:
I compile my document until it stabilizes (All references resolved and such) then edit my document and ...
Is there any way of using the changebars package with XeLaTeX?
The current release of changebars doesn't support a xetex option. Almost three years ago there was a discussion on comp.text.tex about a patch to the package to support XeTeX, but nothing seems to have come of it; there was a simila...
