It is, thanks! Clearly there's some issue in it finding the location of Java, while the OS has no problems. I suppose that's due to registry / file-association info that Windows has.
Having gotten the Strunk & White badge, I can now see my review stats. The trouble is, I see nothing but zeroes both for today, and all.
In the past I've clicked through plenty of questions both in the "First Questions" and "First Answers" tabs on StackOverflow.com/review and voted/edited...
@PauloCereda I'll go copyright it next Monday! :) Then I'll share with you the royalties, provided you send me your credit card number as you'd have done long ago. :)
How do I do a full build with rubber that does the equivalent of
latex foo
bibtex foo
makeindex -s myindex.ist foo
latex foo
latex foo
dvips foo
ps2pdf foo
The files I am using are
foo.tex
myindex.ist
In Is it possible to produce a PDF with un-copyable text?, it was suggested the following code to disable the coyable text from a pdf viewer.
% pdflatex is required
\documentclass{article}
\usepackage[resetfonts]{cmap}
\usepackage{fancyvrb}
\begin{VerbatimOut}{OT1.cmap}
%!PS-Adobe-3.0 Resource-C...
@BrunoLeFloch The CMAP entry for OpenType fonts are inside the font; so unless one is able to find a way to override the resource written in the PDF file when the font is (partially) embedded, I think there's no way. Plus it's a stupid question, as the comments show.
One can defeat DRM on music by playing and recording it; the quality lowers, of course. Text can be retyped quickly without loss of quality, with good proofreading.
LaTeX3 question: in l3regex, should we mention the fact that conversion of a control sequence adds a space, and also what to do to 'unstring' material?
Also, would you object if we made the unsafe functions internal-only: I'm not keen on encouraging anyone else to use these
@JosephWright I'm happy to make the unsafe functions internal only, but some of them are probably used within l3regex, so they should be documented as internal.
@JosephWright I guess that the info about the space after control sequences should go to l3str, and just be briefly mentionned in l3regex. It reminds me that I need to be more careful about the catcode of space in the end-result of replacements.
@JosephWright Yes. I read the question a few hours too late to comment usefully.
@JosephWright We would learn a lot by trying to apply l3regex to something, so it's a good thing that you are doing that. I got side-tracked into writing a proper random number generator, and some arithmetic functions (like factorising expandably, counting divisors, etc).
@BrunoLeFloch I want to check performance against the current siunitx approach, but the regex for what counts as a number is very complex. I don't know how to deal with 'either-or' cases, in any case :-(
@JosephWright Right, you'd need (?:\\pi\ ?|\d). I'm putting \ ? in case the space is not there (but it should be there, I guess). Note that you can use \d for 0|1|2|3|...|9, and ranges are also useful [1-9] matches the same as 1|2|3|4|5|6|7|8|9, much more efficiently.
@JosephWright What do you mean by either-or cases? Is it not covered by alternatives a|b?
@BrunoLeFloch Well, I was working through conversion of user options to regex syntax, so I'd have to do a loop to insert | between each token. (It would have to be \1, ..., in reality.)
@JosephWright Yes. The whole point of storing a regex is to reuse it. I don't see applications where you want to change the regex and give it the same name. But then I have little experience in applied TeX programming.
@BrunoLeFloch I was thinking of the siunitx context. There, I'd want to save a regex based on user input, but would need to change it when options got changed.
@BrunoLeFloch If I copy "sollicitudin" from the PDF into Emacs, I get ^R^N^K^K^H^B^H^S^T^C^H (only the final n is missing); so it's not so clever a DRM, after all.
@BrunoLeFloch Oh goodness, it's complex. Basically everything is optional, but something like 'A comparator (one token) followed by a sign (one token) followed by one or more "digits", a decimal marker (one token), then one or more "digits", then an open-uncertainty token which must be followed ...'
@BrunoLeFloch Ah, useful. I'm not very good at regex construction
@PauloCereda :( "You need to upgrade your Adobe Flash Player to watch this video. " (and I don't have administrative priviledges to the computer I'm using)
@BrunoLeFloch No: r := s = r+1, n := o = n+1, l := k = l - 1. But the same code is associated to the same letter throughout the document, I believe. Some letters are actually not "pasteable", but they are there as something invisible. So a very easy to break code.
@egreg :) I was thinking more along the lines of "what kind of \special and other trickery do you need", I know how to do simple calculations in TeX (programming an expandable lcg random number generator at the moment).
@BrunoLeFloch If you use a CMAP resource in which characters are paired at random, each character will correspond to another uniquely throughout the document, so it's an easily breakable coding anyway: it's just a matter of considering letter frequency and do some guessing. If a text is sufficiently long, it can be decoded fastly.
@JosephWright Enjoy your tea. Not sure what the "comparator" is. For the rest, ([+-]) grabs the sign (add a question mark if the sign is optional). Then (\d+) grabs the one or more digits, and if you want to accept other stuff, you might have to go for (\\pi\ |\d+), or ((?:\\pi\ |\d)+). Then ([.,]?) grabs a period or a comma. In fact, I guess that you want to only allow a decimal part if there is a period or comma, so (?: ([.,]) (\d+) )? For the uncertainty part, (\(\d\d\))? perhaps?
@egreg So I guess the next step in obfuscation is to use tikz (or whatever) to remember the position of each individual character (glyph?) on the page, then produce a pdf with all the characters, but in a (pseudo-)random order.
@BrunoLeFloch Yes, I'd got more or less that far. But then there is the complex part/separated uncertainty and the exponent. I got something working, but it was very hard to read!
@JosephWright You may want to do something like \exp_args:NNx \regex_set:Nn \l_siunitx_regex_tl { \l_siunitx_sign_pattern_str \l_siunitx_mantissa_pattern_str ... } and define each piece separately. Do you think that being able to reuse saved regexes as sub-patterns for another regex would be useful?
@PauloCereda 'tea' = evening meal (my parents are from the north of England - the further north you go, breakfast/dinner/tea becomes more common than breakfast/lunch/dinner)
@BrunoLeFloch Well yes, in the end that would of course be the approach.
@JosephWright But is it useful as explicitly precompiled regexes? The underlying code would be easy (I think), but it raises the question of the syntax to use.
@BrunoLeFloch I suspect my case is rather unusual, as I have a lot of things to match in a flexible way. Most cases would, I suspect, be less complex. For a pure LaTeX3 siunitx-like module, I think I'd like to have a more basic number parser!
@JosephWright I doubt that we can really avoid having a fancy parser. Users of siunitx have come to expect a lot from its power, it wouldn't be good to drop that in LaTeX3.
@JosephWright I can't find any analog of "calling another regex" in the pcre syntax, so if you don't need it, let's not add that possibility. I still need to sort out the {...} quantifier properly. Probably over the weekend.
@egreg That should be enough :). I'm just trying to make my macros robust against spurious overflows.
I have an RNG producing numbers from -(2^{31}-1) to +(2^{31}-1), and I'm trying to produce uniform random integers in arbitrary ranges without throwing away too many.
@egreg Well, really, I'm producing numbers in the [0,65535] range first, then it's just a matter of reorganizing them for any interval (and TeX knows numbers of magnitude up to 2^{31}-1).
One problem is that for very large intervals, computing the length of the interval overflows.
@PauloCereda Well, it might take a little bit long. If you want one billion raw numbers in the range [0,65535], that should take two-three days. (And the period of 2^{32} is large enough that it won't have looped over --- yet.)
@BrunoLeFloch In my last experiment, I left a code running. It was taking a while, so I decided to calculate how much time it should take: something between 10 million years. I gave up. :P
@PauloCereda It's cold enough up here to not be tempted.
@JosephWright Well, I could have a better implementation that wouldn't slow down horribly in those cases, but I'm a little bit too lazy (and hungry, actually).
Basically, the only way for a given code to produce different results when expanded several times in a row is to abuse the silent assignment of \csname undefined_123\endcsname to \relax, and \ifcsname. Since that change is only one way (you can only go from undefined to \relax), I need to consume one csname per random number. Then I provided a non-expandable "flushing" command.
When producing a random number expandably, I need to know which one it is, and go through every csname that was crookedly defined: \undefined_1, \undefined_2, etc., until \undefined_123, until reaching a non-\relax one (checking with \ifcsname).
In principle, I could do that "finding what number the random number has" in logarithmic time: check if the 1 flag is defined, then 2, 4, 8, until reaching an undefined one, then dicothomy. That's harder to implement (and then I need to figure out to do the LFSR computations in logarithmic time).
@BrunoLeFloch If you check in one of Heiko's packages, he harcoded the fibonacci series, wheareas all of us when starting with TeX, went ahead and programmed it. I think there are easier ways to generate them, for what probably would be good for most applications.
In Is it possible to produce a PDF with un-copyable text?, it was suggested the following code to disable the coyable text from a pdf viewer.
% pdflatex is required
\documentclass{article}
\usepackage[resetfonts]{cmap}
\usepackage{fancyvrb}
\begin{VerbatimOut}{OT1.cmap}
%!PS-Adobe-3.0 Resource-C...
@YiannisLazarides I don't think the chinese governement or @Paulo would be happy: they need billions of random numbers, and they won't believe that all of them were 4.
@YiannisLazarides Well, if you can compute 1 billion digits of pi with TeX in less than a week (or even a year for all that matters), I'll buy you a beer.
Any two of these three work fine, but all three break
I now working on an alternative which boxes the content and selects the page size to the box size. Works very nice and simple with pdflatex, lulatex and xelatex, but latex needs that extra PS stuff.
@BrunoLeFloch Well, I'm just scratching the surface. I learned myself some PostScript years ago (Actually before I learned (La)TeX), that helps a great deal understanding such code
@YiannisLazarides Some RNG's tend to exhibit a lattice structure. The chinese probably didn't pick good seeds. (See e.g., random.mat.sbg.ac.at/results/karl/server/node3.html for lattices which appear when using an lcg improperly.)
@MartinScharrer So far, Joseph has done the work on drivers.
@egreg There is a lot to think about, it's true. This worries me, so I try not to think about it. (If I worry too much about the totality of thee job, then I'll never get anywhere. That's before I worry about the suggestion that we're doing everything wrong and should be writing scripts to parse input into XML, then converting XML to primitives, and only then involving TeX.)
@egreg Several layers there. I'd say the driver/low level drawing stuff might be covered by the team, but not the higher level.
@egreg Assuming we get Frank's database-type idea working, then a LaTeX3 class will be very different from a LaTeX2e one
The idea is to set up relationships between things, for example to express 'the gap between a section and an subsection should be <size> if no text intervenes' or 'within a minipage, paragraph indentation should be <size>'
@JosephWright That's very interesting. But I'm always worried about the impact on the greater mass of users: changing syntax for the simple things would mean losing them.
Having gotten the Strunk & White badge, I can now see my review stats. The trouble is, I see nothing but zeroes both for today, and all.
In the past I've clicked through plenty of questions both in the "First Questions" and "First Answers" tabs on StackOverflow.com/review and voted/edited...
How do I do a full build with rubber that does the equivalent of
latex foo
bibtex foo
makeindex -s myindex.ist foo
latex foo
latex foo
dvips foo
ps2pdf foo
The files I am using are
foo.tex
myindex.ist
