@PhelypeOleinik I just added \keys_set:nn { hyp / setup } {colorscheme=phelype} to my code ...

@UlrikeFischer -- Never been a pirate, but in a former life, I used to hang around with some crazy fencers. Have also known a number of reenactors. Learned quite a lot about medieval armory too. Even knew someone whose hobby was fabricating chain mail by hand. (He'd work on it in the same situations I would indulge in knitting, including attending lectures.)

@UlrikeFischer WOOOO, I'M FAMOUS! :)

@PhelypeOleinik ooh

> Brad Cox, Creator of Objective-C Programming Language, Dies At 76

\documentclass{article}

\begin{document}

\input{\sqrt{2}}

\end{document}

@PhelypeOleinik ^

@PauloCereda Breakfast as soon as that document finishes.

@DavidCarlisle oh no

@PauloCereda you may be safer than you think

@DavidCarlisle still...

@PauloCereda we could ask @PhelypeOleinik to fix it....

@DavidCarlisle oh no

@DavidCarlisle we could also fix mathematics

@PauloCereda \input{\sqrt{turing}}

@DavidCarlisle are you suggesting Turing is irrational? :)

@PauloCereda hard to tell which if he doesn't halt.

@DavidCarlisle ooh the Entscheidungsproblem

ooh more Germans

@DavidCarlisle ^^ they had to be involved somehow :)

@PauloCereda I think you mistyped. More serious is the Enteproblem

@DavidCarlisle ooh

@DavidCarlisle Being so constantly corrected by Germans, I suspect it might be Entenproblem :)

@UlrikeFischer @Skillmon ^^ German skill advances. :)

@PauloCereda provably equivalent forms translate.google.com/…

@DavidCarlisle ooh sorry for ever doubting you

@DavidCarlisle no not really. Enteproblem sounds odd. We would add an n. Or switch: Problemente (but it would mean something different)

@DavidCarlisle Why did you try this?

@UlrikeFischer Because David loves breaking stuff? :)

@UlrikeFischer you have mail

ooh mail

@PauloCereda secret github issue mail

@DavidCarlisle ooh

@DavidCarlisle hopefully more productive than the branch discussion thingy :)

@PauloCereda see secret g chat

@DavidCarlisle agreeing to @UlrikeFischer, we wouldn't say Enteproblem. So @PauloCereda is right.

@Skillmon ooh

@PauloCereda the new resident linguistic duck!

@Skillmon nah I got lucky once :)

@Skillmon I trust my sources

Writing documented code is so hard (especially when writing the documentation part weeks after writing the code part....)

@Skillmon write the documentation, prove the specification is correct, then automatically derive the implementation from the documentation. Problem solved.

@DavidCarlisle it's more the implementation comments/documentation part that's missing, not the user interface documentation.

@DavidCarlisle ah yes ;-). Well while it is fitting that you can't input an irrational number, it looks like a bug. I wonder if this is a bug too:

\documentclass{article}

\begin{document}
%\NewDocumentCommand\blub{}{test-input}
\DeclareRobustCommand\blub{test-input}
%\protected\def\blub{test-input}
\input{\blub}

\end{document}

@UlrikeFischer ooh blub

@UlrikeFischer that's a bit like the \frac{1}{2} case it works, but perhaps doesn't input the file you expect

@DavidCarlisle well the question is if \input should do something like \def\protect{}. The error message File protecttest-input.tex' not found. looks rather odd.

@UlrikeFischer I don't get an error message (I have a file of that name) But yes it could do something with \protect but it has to be very late, if you disable \protect too early commands that are made robust will just blow up.

@DavidCarlisle you mean you have a file named "protecttest-input.tex"? That's cheating.

@UlrikeFischer well I also have a file called protect{begingroup1endgroupover2}.tex

@UlrikeFischer and if that didn't have protect at the start would that be a better "filename string" for input of \frac{1}{2} ?

@UlrikeFischer well that's what I meant in the gh issue. there is an automatic error free conversion of \blub to a "filename string" whether or not it's the string you wanted is a matter of personal judgement not a given clear cut rule.

@DavidCarlisle I quite agree with you. In many of this conversions there is no right or false, but only "fits" or "needs adjustment".

@DavidCarlisle I'd expect \input{\sqrt{2}} to do the same as \input{1.41421356}. Doesn't it?

@PhelypeOleinik well perhaps it's working to infinite precision arithmetic and taking a while

@PhelypeOleinik No, it obviously should be the same as \input{√2}. ;)

@MarcelKrüger ooh will there be support for √ANS?

@PhelypeOleinik that's irrational. :)

/sad trombone

@DavidCarlisle Ah, right, my infinite precision parser. I should probably make it stop after one billion digits or so ;-)

@MarcelKrüger That's too exact ;)

@PhelypeOleinik <3

@DavidCarlisle albatross says that I have no Dodo on my machine ;-(.

@UlrikeFischer oh no

On names of expl3 functions: If I create a function that takes as input a token list variable and another N-type argument (a decision function), but doesn't modify the variable but instead leaves the result in the input stream inside of \unexpanded, should it be named \tl_<name>:NN or \tl_<name>:oN? It can't be V, as it requires the variable to expand in one step with \exp_after:wN.

Same problem with a clist (with the resulting clist being left in the input stream, rather than modifying an existing variable), what would be the correct name?

@UlrikeFischer I wouldn't trust that software (@PauloCereda)

@Skillmon :NN

@Skillmon :oN must have a :nN 'parent'

@JosephWright there is \tl_<name>:nN as well. Same for the clist function.

@Skillmon In that case, \tl_<name>:NN is usually equivalent to \tl_<name>:VN, itself a variant of \tl_<name>:nN

@JosephWright but :NN can use o-type expansion, a V variant will also check the variable type and convert them to a tl representation as necessary.

@Skillmon Well yes but that's an internal detail: conceptually, \tl_<name>:NN in these cases is really just a convenience over only providing \tl_<name>:nN then leaving it to the user to create/use \tl_<name>:VN

@Skillmon Point is that if you specify #1 must be a tl var, then the signature is N-type

@JosephWright even if it's just \cs_new_eq:NN \tl_<name>:NN \tl_<name>:oN? :)

@Skillmon Yes

@JosephWright good, thanks :)

@Skillmon Like I say, if you document that the argument must be a variable, then it must be N-type. Or you just document \tl_<name>:nN, provide the o-type variant and leave it at that

@JosephWright you'll see it in a PR in the (hopefully) not so distant future.

Another stupid question regarding expl3: If I add a function that sorts a comma list, but without altering the list, but leaving the result in the input stream (still in clist notation, so that it could, for example, be used inside of a \clist_map_function:nN or similar), in which section should this be put in the documentation?

@PhelypeOleinik not enough z in your version

@DavidCarlisle Oh nozzzzzzzzz

@DavidCarlisle Fixed it for you ;-)

@PhelypeOleinik voted:-)

@DavidCarlisle I feel honoured :)

@PhelypeOleinik not having a German keyboard you could find a z when prompted, it's harder for @UlrikeFischer of course.

@DavidCarlisle I hope she doesn't complain about not enough ß in my answer :)

@PhelypeOleinik but the tick goes to...

@DavidCarlisle show off

@DavidCarlisle Poor OP doesn't know I'm working on ltcmd :)

@DavidCarlisle but the comment is quite wrong -- it has the breaking dependency that you need to be able to find the z.

@UlrikeFischer blub

And now for something completely different: a depressed pig with sheep:

@JosephWright you're free to start nagging :)

Hmmm, this tex.stackexchange.com/questions/580206/… is interesting: latex knows the size of the pictures it is including, so maybe it's possible to tell it to use for example imagemagick to resize them to 150dpi or something like that?

@Rmano sounds complicated, it would make more sense to sent a batchfile over all images to create smaller copies in some folder.

@Rmano I was wondering if you could use the img lib to adjust the resolution on the fly but I don't see anything

@UlrikeFischer yes but for uniformly doing this you should know the size of the picture once included --- and this is quite painful to obtain.

@Rmano why should e.g. imagemagick not be able to know the size of the original picture?

@UlrikeFischer No, I explained myself badly. If you include something as, say, \includegraphic[width=0.5\linewidth]{file.jpg} and you want to have the image in the final document at 150dpi, you need to know the final size of the image to scale it. This info is known just to LaTeX, it's not in the image.

That's basically what gs do when you use -dPDFSETTINGS=/ebook to reduce PDF sizes after-the-fact....

@Rmano well yes, if you want a specific dpi you will have to record somehow their target width, but that you could rather easily write to some file to create a suitable batch file.

@UlrikeFischer yes, that could be an idea. If the log file has something like "including image.jpg at 2.3in x 1.6in" then you can process it, resize the whole lot (once!) and then be happy.

@UlrikeFischer in the case in the question I assume they are all the same width anyway, it's unlikely you choose 5000 individual widths

@DavidCarlisle that's what I do when I prepare the list of students with photos for the labs --- I know the final size and so I resize them all while fetching from the database. But maybe there are cases when the size changes --- hmmm --- maybe include the photos in X tabularx cells? (that would be evil...)

@DavidCarlisle -- Is this in reference to the meta question about blank lines in math? I'm going to look to see if there's a good question about that listed in the math section of "often referenced questions. But I have to run an errand right now. Back in maybe half an hour.

@barbarabeeton no, this is simply a code that loops.

@barbarabeeton no try it:-)