« first day (4827 days earlier)      last day (122 days later) » 

12:03 AM
@UlrikeFischer yes I think the mathml argument needs to be read verbatim, getting the quoting right for the current version is quite tricky but other than that I don't mind much. whatever you add I'll make the examples match.
 
 
8 hours later…
8:05 AM
Some Lua coding to start the day :)
 
8:28 AM
@JosephWright Oh, the joy.
 
8:52 AM
@mickep well a second best if you can't be writing xml
 
@DavidCarlisle Oh no, the hassle...
 
9:34 AM
The French Knuth-Plass talk is now on youtube.com/watch?v=48JNXcaKj2A
2
 
@samcarter Ohh
@samcarter Maybe one can now add translation subs, so one understands what is said as well. :)
 
@mickep French subtitles seem to be available :)
 
@samcarter Autotranslating (can be fun...)
 
9:51 AM
@mickep They should ask David for a flawless translation :)
 
@samcarter We could have a stream event where he translates live.
 
@samcarter Oui, ils devraient
 
@mickep great idea! :P
 
 
3 hours later…
12:56 PM
Hi @WillRobertson
 
1:13 PM
adding the output-directory option to tex was such a bad idea, it adds far more problems than it solves — David Carlisle 2 days ago
^^^ such wise words!
 
@JosephWright hi :-) I see that visiting the chat on my phone is dangerous because the tabs stay active for a long time!
 
@WillRobertson :)
@samcarter That and \outer and \scantokens ;)
 
@JosephWright and \mathchoice and \over ...
 
@UlrikeFischer Not quite as bad though I know what you mean
 
1:31 PM
@JosephWright \mathchoice and \over is worse than \scantokens. \outer can have its useful edge-cases (bm).
 
1:52 PM
@samcarter as always
@UlrikeFischer did we decide on the html format mathml file?
 
@Skillmon \scantokens is basically broken, \outer you could disable and still use TeX just fine ...
 
@JosephWright exactly but \over (and so \mathchoice) is a broken design you just can't really avoid. \Ustack helps a bit but as you can't disable \over it's only opt in so doesn't help much
@JosephWright add scantextokens to classic tex and deprecate scantokens?
 
@DavidCarlisle Certainly worth exploring
@DavidCarlisle Well yes I do see that, but my point was you could drop \outer (as Hans more-or-less does) and it would be fine
 
@JosephWright yes agreed outer can/should be disabled. \scantokens is broken but only affects that one function, \over affects the entire math processing pipleline just to support an infix syntax latex hides anyway
 
@DavidCarlisle I wonder what LuaMetaTeX does: @mickep?
 
2:04 PM
@JosephWright the very first change for 2e was to redefine \newcount and friends not to be outer:-)
 
@DavidCarlisle :)
 
2:26 PM
Shell scripting question: if I loop over files with e.g. for file in *.tex the files are ordered by name. Is there a simple way to have them be ordered by creation timestamp?
 
@DavidCarlisle We didn't decide anything ;-). I just repaired the faulty verbatim-in-tabular in amsldoc (need tagpdf from develop and my newest code ...) and wanted now write something to output a stubs from tex. Syntax is open ...
 
@UlrikeFischer oh would that fix the line endings without me having to patch css into ngpdf's css menu?
@UlrikeFischer let's go with the html format I think it's easier to control white space and quoting around the math in that version and has the advantage of debugging via a browser view
 
@AlanMunn There is no portable way for creation because some file systems and unices do not feature this. On some systems ls will have a proper switch so show it and then you can post-process the ls output. But not a portable sh/bash way.
 
@TeXnician Also VMs can give 'interesting' results
 
@AlanMunn ls -t $(for file in *.tex; do echo $file; done ) (which is filemod not file create but...)
 
2:32 PM
@DavidCarlisle One can always try if stat -c '%W %n' * | sort gives a sorted list matching one's intuition on the system (file birth time is mostly closer than modification, depends if the file system supports it). One can loop through that as well.
 
@TeXnician I suspected that might be the case. For the moment I'll be happy if it works on a Mac, which @DavidCarlisle's suggestion seems to.
 
@AlanMunn for that simple case you can just do ls -t *.tex of course you don't need the nested for. but it depends what you need to do
 
@DavidCarlisle What I'm trying to do is write a script that renames a bunch of files sequentially based on their creation date.
 
@AlanMunn ok but ls -t is modification not creation and if for example it's stuff checked out of git it's more or less arbitrary depending when you checked things locally as git doesn't try to force the filedates to track the upstream repo
 
@UlrikeFischer and @Skillmon -- I asked Knuth about \over vs. \frac. He told me something I already knew -- he chose that method because it sounded most natural to a mathematician. That is, if a mathematician is explaining something to another mathematician on the phone, "over" is what will be spoken, and he will never change. But coding is rather different, and I agree that \frac is easier to cope with; no scoping problems.
3
 
2:40 PM
@barbarabeeton even great men can make fundamental massive design errors.
 
@DavidCarlisle The files are screenshot images that will typically have identical creation and modification timestamps, so that's likely not to be a problem.
 
@AlanMunn you don't use AI to improve your screenshots and remove compromising duck images? shocking.
 
@DavidCarlisle sorry which line endings?
 
The other css I forced on the public ngpdf display for these examples are code lines where verbatim blocks are wrapping

If you look at section 3.3 of the amsldoc pdf you see it renders as

image.png
If I use the sidebar css field to add

span[data-pdf-se-type-original=codeline] {display:block}

then the original codelines are preserved

image.png

I'm not sure if this CSS is something you could do by default or if we should adjust the tagging of verbatim formatted code blocks to change the default derivation to preserve lines?
 
@DavidCarlisle -- I don't disagree.
 
2:47 PM
@DavidCarlisle ah that. It is perhaps possible to add a placement attribute to the code lines, (assuming that ngpdf respect that). I can try. But this isn't what I fixed. You got parent-child warnings and a faulty structure from a verbatim in a p-column where table tagging was disabled and I change the code here.
 
@barbarabeeton infix \over itself would be OK, it doesn't fit latex style and we'd still have \frac but the way it is implemented which means you never know what mathstyle you are in and end up with \mathchoice setting everything 4 times and you not knowing which of the 4 is used is just broken and massively complicates more or less all aspects of handling tex math
@UlrikeFischer ah didn't you add something in to the document for that, you have moved it to the latex-lab code?
 
@JosephWright head something along those lines on the radio this morning
 
@DavidCarlisle :)
@DavidCarlisle I go with 'rhymes with gone'
 
@JosephWright I really want to claim I pronouce cone to rhyme with gone
 
2:52 PM
@DavidCarlisle Ha!
 
@JosephWright we should ask someone with geographically more interesting background such as @AlanMunn
 
@DavidCarlisle :)
 
@DavidCarlisle I didn't read the original tweet. My 'scone' also rhymes with 'gone' but here in the US the things they call scones (which aren't) rhyme with 'cone'.
 
:65012335 I think I do (if I ignore my desire to collapse the chart to a point), although 40 years of being married to someone from the wrong part of the country does mess up your pronunciation biases
 
@DavidCarlisle Not quite a pronunciation battle, but still involving a baked good, is the 'chocolatine' vs 'pain au chocolat' split in the French speaking world.
@DavidCarlisle Given my Scottish parents, the chances of me have the 'cone' pronunciation were nil.
 
 
3 hours later…
5:48 PM
@UlrikeFischer, @mickep Seen the ConTeXt list?
 
@JosephWright For fractions? Essentially as luatex, I think. Or \outer? I do not think it is used.
 
@UlrikeFischer I pushed amsldoc mathml file in html for now, can do the others later after mathml call if it looks useful)
 
@mickep About \over
 
@JosephWright Oh, then Ustack or what it is called.
@JosephWright I don't know why it is so slow :(
 
@mickep Ah, right, yes
@mickep :)
 
5:52 PM
@JosephWright because (unless you break the distribution conditions) the texbook stops at the start and doesn't do anything so is very fast?
 
@DavidCarlisle :)
@DavidCarlisle I think we can assume that the person involved hasn't taken that particular point into account
 
if people used to be happy with 15 minutes per page and now moan about 40 seconds, what would they be doing with all those saved minutes?
 
6:05 PM
@DavidCarlisle -- 15 minutes? I remember twice that on occasion, with something like a telephone book! (3 columns, 40+ lines/column, 9pt type.) If I remember correctly, when it was time to recompile The TeXbook, Knuth would start it up in the evening, and hope it would be close to ready in the morning.
 
@DavidCarlisle This was 40 pages/s. More than what I get...
 
@JosephWright While \over still exists in LuaMetaTeX, it also has \Uover which is basically \frac as a primitive. Therefore you can drop \over completely if you want to.
 
6:23 PM
@MarcelKrüger Hmm, but Hans hasn't removed the_mechanism_, which was @DavidCarlisle's point
 
@JosephWright Sure, but if your format removes the relevant primitives than e.g. \mathchoice is no longer needed since you can rely on \mathstyle providing correct values. (I haven't checked, but I think the \mathstyle bugs are fixed too.) While the complicated mechanism then still exists in the engine, the macro code can basically ignore it. (It's a start...)
 
TIL I can write Mac scripts/apps in JavaScript instead of AppleScript. One less language whose syntax I need to remember.
 
@JosephWright ?
 
6:41 PM
@AlanMunn oh no
 
6:55 PM
@DavidCarlisle ^^
 
@UlrikeFischer ooh
 
@UlrikeFischer ooh
 
@PauloCereda dinner
 
@DavidCarlisle oh no
 
7:10 PM
@DavidCarlisle I added a class display to the code lines and then attached a css to the pdf. You could add also other css instructions in this way:
\begin{filecontents}[overwrite,noheader]{test.css}
.display {display:block}
\end{filecontents}
\ExplSyntaxOn
\pdfdict_put:nnn{l_pdffile/Filespec}{AFRelationship}{/Supplement}
\pdffile_embed_file:nnn{test.css}{}{tag/csstest}
\ExplSyntaxOff
\tagpdfsetup{root-AF={tag/csstest}}
 
@UlrikeFischer ah that's good to know
 
7:28 PM
@PauloCereda Oh no? I take this as an improvement. :)
 
@AlanMunn JS is... interesting :)
@AlanMunn I actually don't know AppleScript, so maybe JS is better. :)
 
@PauloCereda It's actually a pretty nice language I've found. And AppleScript is annoying in that it was designed with non-programmers in mind so the syntax is very wordy and "conversational".
 
@AlanMunn oh no, it sounds like COBOL. :)
 
@UlrikeFischer Speed of LuaMetaTeX
@MarcelKrüger It's still there - I mean that the underlying code has to support it
 
7:56 PM
@DavidCarlisle yes it gives one lots of control. but nevertheless it would be good to know how to tag the verbatim so that it gives lines directly. ngpdf seems to ignore the Placement=Block.
 
@PauloCereda What do you have against COBOL? (former COBOL programmer here)
 
@CarLaTeX nothing in particular. :) I just pointed out that Alan's comment on AppleScript being deisgned with non-programmers in mind is a trait of the design of COBOL itself. :)
 
@PauloCereda ah ecco :D
 
@PauloCereda Yes a bit. But it's been a long time since I wrote a COBOL program.
 
@UlrikeFischer ok see if I get a response to that question but good to know there is a fallback if not
 
8:27 PM
@PauloCereda it seems my initial optimism misguided. It’s so badly documented that it’s almost unusable.
 
9:13 PM
@UlrikeFischer all 5 examples now have html versions of the mathml file
 
9:31 PM
@DavidCarlisle :)
 
 
2 hours later…
11:59 PM
@AlanMunn oh my, that's bad
 

« first day (4827 days earlier)      last day (122 days later) »