@Manuel As @DavidCarlisle says, they are not quite the same. They also date back a long way, to before e-TeX was required. The status of control sequences equal to \relax is slightly tricky in this area (as TeX makes things equal to \relax when using \csname ... \endcsname on an undefined control sequence).
Note her answer is not at all an answer to the original question, but addressed the user’s actual problem. I’ll ask him to change the title of the question.
@SeanAllred This one was particularly impressive. And it took four of us to extract information that would lead to an actual answer (Joseph, me, Akira and finally Ulrike).
I have a question about how to consider macros... so when I store a definition of other macros in a new macro, is it actually replaced by the definition, or is it more so a representation which is loaded into my computer's memory?
Upon calling the macro*, is the macro replaced by the definition
No, I mean that if I store a definition of 500 lines in a macro and call it every time an environment is started, are those 500 lines called every time the environment is used?
Or is it just a representation which is preloaded in the memory at the moment of declaration
@1010011010 it's tokenized but not otherwise compiled in any way, the replacement text as Joseph says is just the list pf macros (and is interpreted with the meaning of those tokens at the point of use not point of definition) (tex is a macro expansion language not a compiler, despite the current fad of calling it a compiler)
@1010011010 I was thinking more the argument of the plain TeX people who don't like anything that hides any primitives from the user (i.e. who hold that the user should always have written all of the macros they use themselves)
@1010011010 everybody hates the language they use and thinks it would be better to use something else, but actually tex's macro language has proved rather robust and usable despite the fact machines now are not comparable to those of 1979, very few languages have survived that transition so well.
@1010011010 it's a question of whether you want \section{foo} in your document or 100 lines of primitices, selecting fonts, spacing and writing tables of contents data
@DavidCarlisle Exactly, so if you build the \section command and you know exactly what the most efficient way is, e.g. how often certain specific definitions reveal themselves in the document, then you're set, right?
The first one avoids extra expansion/token parsing, but I always use the second form as it ends the conditional before working on the outcome. The 'minimalist' approach is only to use the second form if required, but that is very tricky
@1010011010 Yes, but then when your write a different doc you need different code. Depends if you want to redo this every time you write something. That made more sense in the early 1980s (TeX slow) than today (TeX fast).
@SeanAllred When loaded in generic mode, we don't have option handling and I've not set up a similar mechanism. So you always get dvips if you load in generic mode with DVI output TeX.
@JosephWright So is there anything stopping that support from being abstracted away from the LaTeX package option framework, or is just a matter of hammering out the code?
@1010011010 Again, it depends. Someone like wipet would agree with you 100%, and argues that only such cases as are known to occur should be covered by any code, i.e. the code is specific to each doc/use case. I'd prefer to loose some speed in favour of some generality.
@SeanAllred Just a question of specifying some macro that if defined acts as an over-ride. Like I say, no technical reason it's not done but there hasn't been a requirement as-yet. This after all only affects a small subset of box operations at present.
@1010011010 it is the same in any language, for the vast majority of people writing programs in a high level language wins in terms of time to code and document reuse. For a few people using lower level code pays off as they can out-perform the compiler but that is a really small proportion, so some people (here for example) code linear algebra in assembler and have to re-do it every time a new chip comes out with a modified instruction set, but it is a very specialist activity
@DavidCarlisle I've thought about that too and I argued that in the case of a TeX document, a snapshot can prove useful (it is not completely unheard of to compile old documents with dated distributions and/or packages, ok it's not recommended either, but it's a last resort)
In the case of chips and such there's constantly a new one on the market, that instruction set you speak of changes every year and using a dated one directly means you're in a bad spot in comparison to the older releases... seems like less of a problem when it comes to a typesetting language which is more or less stable now anyway...
Not sure what LaTeX 3 will change, though, that is, how will it affect the basic format coming with new distributions
@1010011010 Well that's the plain position very much
@1010011010 If/when we get there, LaTeX3 docs will be distinct from LaTeX2e ones (there is just too much to change behind the scenes). However, we will start from the get-go with a mechanism to define a document date, with the idea being that both the format and packages 'record' changes.
@1010011010 One can bundle up the support files in use for a document, of course (this can be automated)
@1010011010 As you point out, something like wipet's OPmac is a bit odd here as the idea of using plain is often that it's frozen so nothing can change, but I assume OPmac isn't
@1010011010 Yes, for example. I was thinking more a scripting-based approach but the idea is the same.
@JosephWright With added benefit of snapshotting the format (which I assume the snapshot package doesn't do)
I'm not sure how much of OPmac is set in stone but looking at some of wipet's documents it looks like he extends OPmac on a document-by-document basis to meet the needs of that particular document
@1010011010 but if you spend a day (or more likely a week) recoding (say) \section to optimise it for your use it and you are using a big document with say 100 sections, then even if optimistically you save .1 second each run then you need to process your document 10 million or so times before that investment pays off if I got those sums right.
@SeanAllred pdfTeX isn't frozen but (most) development has essentially moved to LuaTeX. There has been some work recently on 'fake' spaces (needed for accessiblity) for example.
"Create an application that would product a PDF output based on templates. The goal is to be able to design the printout using semi-programmable template and apply "data" to the template. [...] We might want to keep the scope small for the initial pass by limiting to simple data elements."
@DavidCarlisle It depends on how perfectionist you are, really. I could easily code a full day to get my version of \section precisely the way I wanted it... not saying I would do it faster in plain TeX (even if I had the knowledge)
@DavidCarlisle Can't incorporate external software when there's any chance of there being a legal problem :) company policy, and a good one for our industry, I think
@1010011010 so it comes down from 10million to a few thousand but it's still most likely lost time and a higher maintenance cost going forward, especially if anyone else ever needs to maintain the document.
The code for pdfTeX is GPL'd :( This might make things difficult to swallow. How do you go about explaining that even if a program is GPL'd, its use is free?
@1010011010 if a document never changes you may as well just keep the pdf. But any document that is maintained changes due to circumstances or technology, so someone wants an ebook or whatever replaces ebooks in 2020, if you use standard document markup chances re there are standard tools to do whatever, if you have just some one-off macro set that works in one document, that no one else knows how to extend or convert, the document is far less useful.
@JosephWright Well, when I sent in the docs to our legal team requesting permission to work on AUCTeX, they denied it. (I'm going to fight that when I have more time.)
@SeanAllred Classic corporate situation by the sound of things: paid-for software is fine even though the license may be restrictive, free software no good as there is no payment/terms/...
@SeanAllred I worked freelance as tex consultant for a while (9 months or so) after leaving University post doc, before starting here, didn't like it much:-)
@SeanAllred apparently you are supposed to think of random amounts of money and charge people and stuff, I suspected I wasn't doing it right when clients were telling me I wasn't charging enough...
The latter give talks at BachoTeX, demonstrating their rather impressive 500- or 1000-page catalogues of random things, in 10 or 12 different languages.
@SeanAllred you know I've not actually seen it since about 1995:-) in those days there were not really any "distributions" as such you got TeX on a big tape and compiled it... As far as I recall it's as you expect a separate tex executable hidden in the background running wen you ask to print, they also had tie up to maple (later I think mupad) working in a similar way.
@ArthurReutenauer well in SW's case it was more the other way round, you wrote the document using tex syntax in the editor but it would get maple to evaluate expressions
@SeanAllred but also probably didn't do what they needed eg cutting and pasting images from the windows clipboard means you really want windows metafile support in the driver (actually i saw a question asking about that the other day..)
@SeanAllred It starts here: sourceforge.net/p/xetex/bugs/54 where Martin Schröder points out, correctly, that since one source file of XeTeX uses poppler, that is under the GPL, the whole of XeTeX should be relicenced under the GPL.
(Martin would know, he followed the development of free-as-in-speech PDF libraries pretty closely at some point, and the offending source file was to read with reading PDF files.)
Note the report is four years old, but I only became aware of it last week.
Problem with relicencing XeTeX is of course that we would have to chase the copyright holders, one of which is SIL International, Jonathan’s employer up to 2008.
@SeanAllred Well I only became involved in XeTeX one month ago, so I plead not guilty. But generally speaking yes, it’s not good.
Anyway, I was not looking forward to having one more licence discussion - it happens regularly with hyph-utf8 and it’s always awkward to have to contact copyright holders who haven’t been active in 20 years, etc.
Peter Breitenlohner and the binary builders may start to hate me, but I can live with that (it’s either that or a gazillion angry XeTeX users, so I’m choosing the former).
@SeanAllred I wonder if for pdfTeX it’s not a policy choice. But I don’t know for sure.
@SeanAllred I bought the domain name XƎTEX.net one night when I was drunk (many years ago, when I had just started using XeTeX), I may make use of it now ;-) It’s parked for the moment.
@SeanAllred Actually the sources for xdvipdfmx are in the TeX Live repository, and alas, I just checked that xdvipdfmx is under the GPL (wasn’t aware of it).
U:\>dvipdfmx --version
This is dvipdfmx Version 20150315 by the DVIPDFMx project team,
modified for TeX Live,
an extended version of dvipdfm-0.13.2c developed by Mark A. Wicks.
Copyright (C) 2002-2015 the DVIPDFMx project team
Copyright (C) 2006 SIL International.
This is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 2 of the License, or
(at your option) any later version.
I'm still going to try and push this through. TeX in general is unbelievably important to me and it would mean a lot to me personally to be able to use it in some small way.
@SeanAllred As Joseph pointed out, dvipdfmx, on which xdvipdfmx is (obviously) based, was already under the GPL, and it’s very much the core of the program (again, I suppose that’s obvious), even though the extended DVI file support is an important part too.
Hi, is there any “default”/“known” way of avoiding a page break between “Proof.” and an enumerate/itemize. The code I have is \begin{proof} \leavevmode \begin{enumerate} ... Any fast suggestion (I can't find anything on TeX.SX, but my bad relationship with the search engine comes from long time ago.)
@SeanAllred Well, feel free, if that would be useful to you. I’m happy to be kept in the loop, but don’t expect me to contribute to the discussion too much, I’d rather spend my time on real things :-P
@Manuel no as I say it depends how you define proof but if you have \leavevmode all bets are off as then you have a blank all white paragraph between the heading and the list and chances are anything the heading does to suppress a page break will only apply to the blank paragraph.
@DavidCarlisle Well, then my question is, what is the preferred way of starting a proof (or any theorem-like environment) with an enumerate or family without the first item being on the same line as the “Proof.” (i.e., not indented).
@SeanAllred really even in a commercial environment GPL shouldn't be so much of a problem (an awful lot of commercial code is compiled with gcc for example) it would be bad for us as it would limit the use of a library but it is hard to see how you would be extending xetex or xdvipdfmx programs and so GPL has few restrictions (although whether corporate policy allows that is another matter of course)
@SeanAllred One of the most horrifying licence discussions was when Mojca was contacted in 2010 about one pattern file in hyph-utf8 that was in the public domain. The reason why that’s potentially a problem is because in most countries, there is apparently no mechanism to voluntarily place one’s intellectual property in the public domain - with the notable exception of the US and the UK (admittedly a rather large exception).
@SeanAllred So some Linux distributions apparently refuse “public domain” as a valid status.
@SeanAllred but that's what I mean, do you really need to compile xetex from source to make a derivative work (GPL conditions all over) or just call xetex as a program to print stuff (probably unrestricted if you do it with care)
@SeanAllred That’s what the CC0 is for - I suppose that legally that means you’re still the copyright holder, but since you waive all responsibility you’re pretty much safe from lawsuits if anything goes wrong with your work - one would hope.
@SeanAllred Isn’t it? :-) I guess you could try and make the point that you’re actually a zombie and have been for 70 or whatever the requisite number of years is, but I’m not sure that would hold in court.
@SeanAllred The applications I'm thinking of have all provided their own drivers, so that was not an issue. In recent years, of course, there has been a tendency toward a smaller number of drivers, all as you note GPL
Anyway, the OP had a document from 1994 and absolutely refused to update it since he expected to still be compilable (it was late 2007 or early 2008 when he asked the question). There were about 3 or 4 specials for including graphics (using dvitops’ specials syntax), so obviously I suggested to change them to use \includegraphics instead.
But the OP didn’t want to, so, since the mention of dvitops had aroused my curiosity, I got the sources from CTAN and managed to compile it with gcc and a minimal amount of tweak to the sources.
I then went back to the list to explain how simple it was to compile dvitops and went ahead to describe my hacks.
This had the desired effect: the OP gave up on using dvitops.
@DavidCarlisle Thanks, I think I noticed before, about Saxon. Actually I don’t have much use for it anyway, the only time I tried out XSL stylesheets was for UDHR in Unicode that are 2.0, but it wasn’t too much work to make them into 1.0
Incidentally, XƎTEX.net is now live (for me) and redirects to Sourceforge.
@CBenni with r not R then it should not move at all. Do you hav e the paragraph starting immediately after \end{wrapfig} ? and of course what's in the input file?
@CBenni wrapfig allows {r} for right aligned at current spot, or {R} for right aligned, floating, so as you use r the figure should not move away at all, never mind 6 pages.
@CBenni a proof environment is (in most definitions) a list and the first rule of wrapfig is, don't put it in or near a list.
- You must not specify a wrapfigure in any type of list environment or
or immediately before or immediately after one. It is OK to follow
a list if there is a blank line ("\par") in between.
@CBenni follow the instructions in wrapfig.sty :-)
@CBenni A bug is when software says it does something and then doesn't. When the documentation says that you should not do something, but you do, that's "user error".
@CBenni It's always possible to achieve any effect, but perhaps not with wrapfig, for modifying list shapes it's tricky to do something in general but something is usually possible, depends a lot on the details of the proof environment and the text that is being wrapped though.
@CBenni first hit for "wrap text in theorem" is this, there were loads of others, something probably does your use case:
I want to illustrate my examples, which are set in a theorem-like environment. Is it possible to wrap the text and equations around the figures? I have read about several packages but I can't get them to work in a theorem. Am I doing something wrong?
yeah ok whatever, too much hassle. remove the theorem environment for that proof and faked it to look the same. I am trying to get work done here and not experiment with new packages. Theres a deadline to meet. Thanks anyways @DavidCarlisle
I dont quite understand the design decision "you cant use it near lists." however - to me, that sounds like "we were unable to get it to work, so were just not gonna support that"
@CBenni It's not a design decision, it's what could be achieved in a reasonable amount of code.
@CBenni yes except not "we" just Donald Arseneau, who's one of the best TeX practitioners on the planet, latex reshapes paragraphs to add the item labels and locally over-rides almost all paragraph related primitives (including \par) in order to do that, so having something that also reshapes paragraphs to add the image cutout and works seamlessly across list and non list material is decidedly non trivial.
@egreg you lost the tick but you did have the right answer
your redefinitions lose the automatic italic correction added by the default definitions you should use \DeclareTextFontCommand as shown in egreg's answer — David Carlisle1 min ago
I didn't find a question/answer on the site about this: is there a way to tell Beamer to put the slides in an order that is different than the order they appear in the .tex file? I tried putting them in my preferred order in the \includeonlyframes{} list but they come out in the order they are found in the file
@egreg Alan I think, can't remember, I suspect it was Frank/Chris's suggestion (goes back a long way to the original 2e sketches)
@ArthurReutenauer \parshape only applies to one paragraph so if you need to cut out 4 lines and the paragraph only has 2, wrapfig needs to catch \par and construct a new shorter \parshape for the next paragraph to hold the bottom of the image, trouble is list environments use exactly the same trick of using \par and \everypar to propagate a \parshape....
@ArthurReutenauer Frank said to me once that xgalley is more complex conceptually than xor. I'm not sure I guy that, but it is certainly a tricky beast.
@ArthurReutenauer lists are one of the more complicated latex structures, there's a reason Leslie relaxed after they were defined and just defined everything else to be a list: center, verbatim, quote, ....
I want to illustrate my examples, which are set in a theorem-like environment. Is it possible to wrap the text and equations around the figures? I have read about several packages but I can't get them to work in a theorem. Am I doing something wrong?
