@samcarter ^^^^^^ what do you think of this proof? Is it detailed enough? ;-)

@CarLaTeX are you there?

@manooooh Here I am

@CarLaTeX hey :)

@CarLaTeX do you want to represent wonderful words?

@manooooh Is it possible to play on the smartphone? I'm not at home at the moment

@CarLaTeX you have to install the official app :/

@manooooh Let's postpone to tomorrow, then

@CarLaTeX ok!! Best of luck :)

@manooooh Ciao

@UlrikeFischer Cool

@UlrikeFischer Let me know if/when I should set up a fork in the LaTeX3 area

@UlrikeFischer On the file renaming business, perhaps I should add a generic 'rename these files' idea to l3build

@UlrikeFischer I've had an idea of how to handle multiple configurations and tests, so the .diff files don't get zapped. I'll implement today if I get time.

@marmot The Riemann Duckjecture?

Can somebody who is using MikTeX look whether ducksay is updated to version 2? And if so, whether the files ducksay.animals.tex, ducksay.code.v1.tex and ducksay.code.v2.tex are included?

In TeXLive the three files are not included which makes the package unuseable :(

I think I have to write one of those *.ins files (which I don't understand yet)...

@Skillmon Where's the source? Is there a repo we can check?

@JosephWright github.com/Skillmon/ltx_ducksay

@Skillmon They end up in the doc tree. Write on the TeX Live list about it.

@egreg thanks for pointing out. What can I do to prevent such issues in the future?

@Skillmon Can't help noticing your licensing. I'd avoid adding any clauses to a license: it's not a good long-term plan. If you don't like GPLv3, don't pick it ...

@Skillmon I see it happening rather frequently when there are auxiliary tex files.

@Skillmon i see no update yet. When I get it I will check if it is complete.

@JosephWright lets wait a bit. I just planted in various place my address and three more or less official locations would be a bit too much imho.

@UlrikeFischer Sure: I was also thinking a proper 'clean up' is going to be required

@JosephWright I do like the GPLv3, it's just not the common license for a LaTeX package. That's why I also licensed it under the LPPL1.3c. Maybe I'll drop the GPLv3 in future releases. My other published packages are under the LPPL only.

@Skillmon I meant the clause you've added to GPLv3 ...

@Skillmon We (team) have been discussing dual licensing recently :)

@JosephWright "This package is distributed under the terms of GPLv3 or later, or the LPPL 1.3c or later, which ever license fits your needs the best." No clause added, or did I formulate it wrongly?

@JosephWright Oh yes. E.g. there are at least five locations where the various involved file lists are stored -- with varying syntax and naming scheme (sometimes with prefix, sometimes not). The version infos are different everywhere, which makes tagging up to impossible. There are lots of tex-files which probably should be in the docs and more. I will made a list of issues ;-)

@Skillmon I wasn't questioning any particularly license

@Skillmon You have a 'rename' clause

% Alternatively you can use this work under the terms of the GNU General Public
% Licensce (GPL), version 3 or later, except that you might not redistribute it
% under the same  name if you change  its functionality.  Choose which of these
% licenses fits your needs the best.

@Skillmon ^^^

@UlrikeFischer Cool

@UlrikeFischer What I was imagining is setting up a 'new' luaotfload repo in the LaTeX3 GitHub organisation, not forking from an existing one but rather pushing from e.g. my laptop (so breaking the 'link'). Then asking the other 'owners' to remove their ones, so we have a single definitive place. First, though, branches and issues do need cleaning up.

@JosephWright then I'll drop the GPL in future releases... Doesn't hurt.

@JosephWright did you saw my last messages yesterday? The documentation \inputs a file and I had a really hard time to copy this to the doc folder, both for compilation and for ctan and had to set supportdir at the end, which doesn't feels quite right. Could you take a look at build.lua and tell me how to correctly include luaotfload-main.tex?

@UlrikeFischer I missed the message. I'll make myself a fork and take a look (so I can send a PR with fixes)

@UlrikeFischer If the easiest fix is moving a few source files around, will that be OK?

@JosephWright lualibs and luaotfload are under GNU GPLv2 license exactly (from context). Would this be a problem for a maintenance in latex?

@UlrikeFischer One for an email :)

@UlrikeFischer I'll reply shortly

@JosephWright depends on which files. Moving the files in doc should be okay. But there is no complicated structure in the docs: almost everything is in ./doc with the exception of two text files in the main dir.

@JosephWright You mean to replace the various import scripts to get the files from context? I'm not sure if this is very generic code ;-).

@UlrikeFischer No, I meant for the 'two README' situation. I could code in a specific 'name of README for CTAN' variable

@UlrikeFischer About to have a look: got sidetracked

@JosephWright ah yes. That would a good idea. Currently github shows it as soon as one switch to the doc folder.

@UlrikeFischer I'll think about it: perhaps one for email

@UlrikeFischer Do you have anything non-standard in your set up? For me, l3build doc on a fresh check-out doesn't work cleanly

@JosephWright well the texmf tree with the files must be active (luaotfload-tool e.g. must find its script, and also the new lualibs files must be found). On travis the tree is set as texmfhome. the texmf tree is build with l3build install. the build lua sets the option for this relative location.

@UlrikeFischer Ah, right, so it is non-standard: explains things

@UlrikeFischer So you export a custom TEXMFHOME before running the scripts?

@JosephWright only on travis. locally the texmf is simply one more root added with tlmgr conf auxtrees.

@UlrikeFischer Yeah, but that's still non-standard

@UlrikeFischer The 'standard' team set up would be to have all of those files in the support dir, and copy them across ... but it's your call, of course

@JosephWright if you can tell me how in the standard setup luaotfload-tool.exe can use luaotfload-tool.lua ...

@UlrikeFischer Hmm, the tool part is tricky ... I'll think about it

@UlrikeFischer I'll need a while to fiddle about, and perhaps adjust l3build. Give me until this evening?

@Skillmon note that a clause saying you had to rename was what got LPPL classified as non-free by Debian, and why LPPL 1.3 drops that (and just encourages renaming without enforcing it)

@JosephWright it is not only the tool party. for some tests involving kpse the result is different if support files are in the current directory or in a texmf. E.g. in the lua-font-pond I'm testing if lua fonts are found in fonts/misc/luafont.

@UlrikeFischer Hmm: the results shouldn't be variable in the first place, and should be normalised otherwise

@UlrikeFischer At the very least, I'll need to put something into l3build such that ther's no need to set anything other than 'have a TeX system installed'

@JosephWright what I mean that I have to test this feature. files in the current directory are normally found always, but files searched with kpse only if the code (the kpse option) is correct.

@UlrikeFischer Hmm: files in the local directory should only be found if the appropriate input is activated ...

@UlrikeFischer can't you arrange that there is a local texmf.cnf in the test area which sets up whatever kpathse paths you need

@DavidCarlisle There is that I guess

@CarLaTeX @egreg I just had the most gruesome lunch in years: My wife ordered food as we didn't have much time to cook today and thought she might do the kids a favour by ordering hotdog pizza. So my lunch consisted out of a mixture of Hawaiian pizza and hotdog pizza (with way too much ketchup -- who puts ketchup on pizza anyway?).

@DavidCarlisle Hm, yes. should be possible. I would have to write it on-the-fly so that it works on different systems. I will think about it.

@JosephWright what I meant is, that some parts of the code related to searching of font files can only be tested if there is a real texmf tree around.

@UlrikeFischer well yes that's sort of why the tu-config tests in latex2e/base are separate, then main test suite is in a sandbox that doesn't use the installed files but the TU tests need to get (at least) the fonts from the local system

@JosephWright If I setup such a local texmf tree, can (and how) then prevent l3build from copying the/some of the source files to the current directory?

@UlrikeFischer Like I say, I guess I need to think about the detail here: this is a complex situation

@DavidCarlisle but I don't need only access to my local system, but to dedicated files in a texmf tree of the repo. As an example look at the "font" files here: github.com/u-fischer/lua-font-pond/tree/master/texmf/fonts/misc/…. There are loaded in github.com/u-fischer/lua-font-pond/blob/master/testfiles/… with a kpse search.

@UlrikeFischer In the first instance: exactly what command do you run at the moment (I want to start from a known place!)

@JosephWright for which step of the process?

@UlrikeFischer Well, whatever you do before you run l3build doc or l3build ctan or whatever for luaotfload. I want to be sure I start with something working!

@UlrikeFischer yes sure but if the tu-testfiles area had a texmf.cnf pointing at a local texmf tree that would do the same (but Joseph might give you an l3build setting for that anyway once it's clear what the spec is:-)

@DavidCarlisle Exactly: if we need new features here, I'll supply them

@JosephWright from current the state of the repo, adding the texmf tree with `tlmgr conf auxtrees add path/to/texmf" should be enough. Then l3build doc should work. But @DavidCarlisle is quite right, using a local texmf.cnf would be better.

@UlrikeFischer Right, I'll try that

@UlrikeFischer I don't like the sound of l3build doing tlmgr conf as that's changing the default setup on the local system (unless I misunderstand?) (@JosephWright)

@DavidCarlisle Yup: I've just been stopped as it needs Admin rights :)

@JosephWright on my machine it doesn't as I installed the "system" texlive as me, but I don't think an l3build check should be changing the tex paths on the system that have an effect for later normal documents (and you wouldn't be able to ensure "undoing it" at the end of the test run in cases of failure

@Skillmon Anathema!

@DavidCarlisle All true

@DavidCarlisle i just said that a local texmf.cnf would be better ;-).

@UlrikeFischer I think we all agree on this: doesn't need new build features, is self-contained, ...

@JosephWright ok. Make a suggestion ;-). If it could also at the same time setup the location for l3build install it would be even better.

@UlrikeFischer Shouldn't need to install files for testing ... that's rather the point of l3build

@JosephWright back to square one. How can luaotfload-tool find its script?

@UlrikeFischer I'm thinking :)

@UlrikeFischer Like I said, I need to do some tests: might take a while

Cannot resist. From where did this room originate? I congratulate myself as one of the mystics of TeX/LaTeX.

And Mathjax.

@humn Huh?

@humn The TeX site ...

I learned from Knuth.

And then. hang on.

@humn Knuth himself showed you this chat?

Donald (not Trump) taught me secrets not yet revealed.

And from David for MathJaX. And i taught each a thing or two as well.

@egreg Precisely

Neither brought me here. I'm just nosy.

Can't wait to share my \secrets!

Sorry, i should backlog to find out what has already been shared.

@UlrikeFischer I think I might need to set TEXMFCNF in the script, then making a custom texmf.cnf should work ... thinking

Knuth had one goof, or so admitted. "\loop"

I think it's brilliant but he is embarassed.

It has a bail-out, which every loop needs.

Sorry to but in. When i saw a site for (La)TeX, the fingers went here.

I make packages in both TeX and LaTeX in my sleep. And MathJaX.

Just can't help it.

Please welcome me.

Or not. I will continue.

@humn Hello

@JosephWright , thank you!

Anyone who can \understand this language is truly brave.

Next, PostScript. Don't get me started.

@humn PostScript is just stack based

@humn TeX is perfectly clear

Each is perfect in their own ways.

PS is pure. (La)TeX is truly inspired.

I've never known a language as pure as PostScript. What's it called now? PDF?

But the backslash grammar of where we are is a work of art.

A labor of love. (From one who knows the lover.)

@JosephWright, ya still here? To say PS is stack based is to understand. Welcome, stranger!

PS is backward strung while (La)TeX is forward strung.

In my honest opinion.

@JosephWright yes, that's actually how I'm adding roots currently (my system predates conf auxtrees. but context doesn't like such local texmfcnf settings, so sometimes one has to reset texmfcnf ;.).

@UlrikeFischer , please let me interject. Folks who dare \you-know take more time to think than i. And i'm one those folks. But i don't. Hello!

You make perfect sense and i wish i knew the solution.

Someone does.

That's a taunt.

@Joseph and @Ulrike, you are lucky i'm on your side(s). Battle for good!

When i'm not online i'm onstreet. For good.

That got sidetracked. Again, you have not met a \TeX master as i. Let me boast.

I make pictures beyond description with TeX, just because i can.

@UlrikeFischer OK, setting up a texmf.cnf file 'manually' gets the tests to pass: I'll integrate

I'm trying to post one of the most beautiful pictures ever. But the system is stymeying me. Perhaps for the better. Stay tuned.

Sorry, it ain't working. I'm gonna take a spell and come back with all barrels blasting.

Bye for now.

@Skillmon Horrifying :):):)