@JosephWright I'm not sure about the naming (and the setup). I mean you are not really adding a texmf dir but simply a folder and all subfolders which can but doesn't have to follow the TDS. This could have surprising side effects for the seach pathes if there is a real texmf present.
@JosephWright and also metafont (in theory) hence texmf .... I think "texmfdir" is OK (that doesn't imply "TDS" historically "texmf" predates the TDS layout specification)
@JosephWright @UlrikeFischer speaking of directory layouts am I right in thinking that if I want to force ctan.org/tex-archive/macros/latex/contrib/hyperref with a doc subdirectory on ctan then I need flatten=false and have that layout in the source (which actually, I do) and we don't have anything like the tdslocations variable to force a specific ctan layout? ...
@JosephWright we have to go now. I can live with the name, but the docu should make the difference clear and perhaps it shouldn't be activated by default so that it doesn't add search pathes in existing setups.
... Also to get the html manual I need to call htlatex which I can do from build.lua os.execute or whatever, but I don't think we have any direct way from within l3build?
@UlrikeFischer I've adjusted so it only gets used at all if the variable and the dir exist
@UlrikeFischer I'll make a doc fix: can go with the 'automated deploy' tests
@DavidCarlisle Yes, that works: see the beamer set up for example
@DavidCarlisle At present that would need some hook adding: you can specify manual replacements for the std typeset target, etc., so that would be the place to look
@DavidCarlisle, @UlrikeFischer I'm going to do the deployment stuff for l3build then probably the PRs
@JosephWright yes I got flatten working yesterday but I wondered if a ctanlocations list in the style of tdslocations would be more natural than just having a boolean choice of flat or mirror the source layout. (not urgent...)
@DavidCarlisle I've said for a long time that I'm almost certainly their biggest 'customer': I can't really imagine anyone else sends as many updates as me
@DavidCarlisle But they know we are using l3build, and basically nowadays that means the updates are just right
@PauloCereda Yes, it's looking good: @MarcelKrüger did the set up work for luaotfload, I've just copied it. But it is actually very easy to use
@JosephWright other people have wondered that too:-)
@JosephWright oh that's odd I swear I signed up before and got a message saying I'd be notified if I got access, but I just tried again and got instant access.
@JosephWright ah I see that the "old" actions syntax will be discontinued tomorrow, so I guess I just never got access to that limited trial but the new version has a wider beta.
Please use an explicit test - I know gcc suggest just an extra set of
parenthesis, but I'm personally convinced that is just because some gcc
people have been damaged by too much LISP.
- Linus Torvalds discussing gcc requirements on linux-kernel
@DavidCarlisle No. I only meant that you use Emacs Lisp, but probably don't use any other Lisp. Though it's likely to be valid syntax across the Lisp family.
But my knowledge of Lisp was never more than beginner, and is not quite faded, since I haven't looked at it in a while. Though I would like to.
@FaheemMitha I have used dsssl (which uses scheme syntax) quite a bit and some other ones off and on over the years, did try a few common lisp examples but never really had a project that needed that
@DavidCarlisle I see. I didn't mean to suggest you don't know about Lisp, just that you don't use it. Few do, unfortunately, outside the hard-core Lisp community. The sort of people who hang out on #lisp in Freenode, for example. Or blog about it.
@FaheemMitha probably using lisp syntax is what killed dsssl and xslt (using xml syntax) took over completely. Of course move on a couple of decades and main reason people now give for not liking xslt is the xml syntax. Fashions change....
@DavidCarlisle It's called set! in Scheme, setq in Common Lisp. I think the choice of the name in Scheme is to call attention to the imperative construct.
And in dsssl it did not exist, as dsssl used a purely functional subset of scheme.
@HaraldHanche-Olsen so it is, I did actually check before the comment but I was mislead by the example I landed on as someone had defined a local setq syntax in scheme based on set! :-)
@HaraldHanche-Olsen yes an extended subset so not really scheme at all, so scheme people didn't like it and people who didn't like lisp didn't like it and everybody else thought it was lovely:-)
@HaraldHanche-Olsen well given that I searched for setq scheme I suppose the search would be biased towards pages where there was something called setq defined...
@FaheemMitha That's what James Clark thought when he persuaded the SGML style language group to use scheme lisp syntax father than invent a new custom syntax but in reality the number of people who find (+ 1 2 3) easier to understand than 1+2+3 or sum(1,2,3) is .... small.
A fixation on the superficial is the primrose path to mediocrity.
@DavidCarlisle In reality there is nothing to understand. It's just notation. Learning to drive a car without killing people is substantially harder, for example.
@FaheemMitha it might be better to have a mediocre system that people use than a brilliantly constructed system that no one uses. DSSSL was perilously close to being the latter.
@FaheemMitha no one understood it, the one implementation was good but incomplete and there were never enough users to motivate more implementations or to complete that one. Then the XML hype hit and the same group of people had the (excellent as it turns out) idea to salvage the best ideas but to reform them in a new much simpler language embedded firmly in the XML world, dropping any SGML baggage, and dropping lisp syntax.
@barbarabeeton I am sure the eventual latex3 format will have catcode j by default so Christmas is still secure.
> One touch of nature makes the whole world kin, That all with one consent praise new-born gauds, Though they are made and moulded of things past, And give to dust that is a little gilt More laud than gilt o'er-dusted. The present eye praises the present object.
Though PHP isn't even gilded. It looks like what it is. Excrement.
Or, the short form, courtesy of Buffy the Vampire Slayer - ooh, shiny.
@FaheemMitha you see you are very inconsistent, with lisp that you like you say syntax is not important and query why that should put people off, but for xml and php that you don't like, you comment negatively on what they look like, that is, on their surface syntax.
@DavidCarlisle Well, PHP is just terrible all the way through. So, it looks like what it is. And I admit I just dislike the look of XML, though perhaps I'm wrong in doing so.
But I never disliked the look of Lisp. Of course, there was once a time I didn't understand what it meant.
It's possible XML syntax is also sensible syntax. But I've never used it for anything.
I try to keep an open mind. But everyone has his/her biases. But one should try to stay alert to them.
@FaheemMitha I use it for more or less everything, it also underlies the object model for this page of course. For document oriented structures it is far more natural than lisp to use <p>stuff in a para</p> than (para-constructor 'stuff in a para') but of course either way it's just an annotated tree and you can convert between them without loss.
@FaheemMitha so people like to use <a href="http://ww.example.com">an example link</a> but you could just as well encode it in xml as <element><name>a<name><attribute><name>href</name></attribute><value>an example link</value></element> and the latter is trivially converted to s-expression syntax, it just depends if you want to highlight the tree structure or the view of marking up a document with more importance given to the text than the annotations.
@DavidCarlisle I have to admit though, every time I contemplate xml namespaces I get a headache. Looks like a bad example of design by committee to me.
@FaheemMitha look what you wrote above "It's just notation" You are free to prefer lisp s expressions many orders of magnitude more people prefer to use <a href="http://ww.example.com">an example link</a>
@FaheemMitha Just like with Lisp syntax, with good editor support it should be possible to tone down the tags so you can concentrate on structure and contents. In my emacs setup, I have parentheses i Lisp shown in a very faded colour. Instead, I rely on indentation to see the structure.
@FaheemMitha why? do you think \begin{enumerate} ...\end{enumerate} is distracting? That is trivially equivalent to <enumerate>...</enumerate> or (list ....)
@FaheemMitha Yes. Now I don't write or edit much xml, so I don't really know what can be done in terms of (partially) hiding the tags. But it would be a priority if I did.
@FaheemMitha latex \begin{enumerate} \item aa \item bbb \end{enumerate} requires a custom parser to parse, and even using that parser detecting the end of an item, eg to add special formatting, is tricky. the xml (xhtml) equivalent <ol><li>aaa</li> <li>bbb</li> </ol> has so many advantages, it is trivially parsed by any number of off the shelf parsers, finding the extend of ever element is trivial, and it is far more easily used for multiple purposes.
@FaheemMitha yes take this page for example, and choose "view source" from the browser menu. What you see is a lot like XML and not at all like lisp s expressions.
@FaheemMitha you wrote I'm not sure what you mean. I've written web pages before and clicked the send button that wrote that for you. But there is a reason that it wrote it in that form, not encoded the equivalent tree as a lisp expression (or JSON)
@FaheemMitha no it is not gobledegook it is structured annotated tree in a clear format so people can and do use that in multiple ways. I didn't like the site restyle so I use a user css file that restyles that, @PauloCereda wrote the Psmith bot that read such things and answered questions contained within it, some people use a use javascript to detect latex syntax and inject mathjax, etc.
@FaheemMitha you could encode it in latex, replace <foo> by \begin{foo} and </foo> by \end{foo} it would be equivalent but much harder to find a parser or to resuse the data in any way.
@FaheemMitha no but you asked why dsssl using lisp less popular than xslt using xml syntax, well to style something for the web you need to generate things like that, and in xslt to generate <span class="meta"> you write <span class="meta"> in dsssl you write something like (I forget the function names) (generate-flow-object "span" (add-property "class" "meta"))) and oddly enough people find the xslt version easier.
With all the stuff that's going on, it really makes me think of the proverb. You know, may you live in interesting times. I'd say these times are interesting. Wouldn't you agree?
Just wondering, how many people here are aware of what's going on in Kashmir?
For reasons that are unclear, AUCTeX thinks that a document is resulting in a 0 page PDF, even though it's actually 1 page. I suppose I should report it, if I can work up a MWE.
When I try to compile the code for this question or its answers, I have an error. A year ago, I didn't have that mistake.
Package xkeyval Error: language' undefined in familiesglobal'.
\documentclass[12pt]{article}
%\usepackage[utf8]{inputenc}
%\usepackage[upright]{fourier}
\usepackage{alte...
@UlrikeFischer you can increase the number so you get them all on one page, eg ctan.org/… then curl or wget gets that as a file, and you can clean up the html as needed?
@UlrikeFischer -- Send a suggestion to Gerd Neugebauer. He'd surely know (if it's possible) or have an opinion as to whether it might be found generally useful.
When I try to compile the code for this question or its answers, I have an error. A year ago, I didn't have that mistake.
Package xkeyval Error: language' undefined in familiesglobal'.
\documentclass[12pt]{article}
%\usepackage[utf8]{inputenc}
%\usepackage[upright]{fourier}
\usepackage{alte...
