Am I having a strange bug or is it the way american write percent ? imgur.com/a/HoMXqh4

@Matheod That's the font you're using

@Matheod For example here, one of the differences between Times Roman and Times New Roman is that thingy in the % sign

Thanks. Who got this stupid idea to add this curve ><

@Matheod -- That's the way I was taught to write it. Its appearance in particular fonts may differ. It appears with the "extra" upper stroke in the Unicode charts (U+0025): unicode.org/charts/PDF/U0000.pdf

@HenriMenke space factor, I think … since the expansion of \TeX ends with an uppercase letter.

@HaraldHanche-Olsen Correct.

'ello

@PauloCereda 'inner

@DavidCarlisle oh no

/quacks in despair

@DavidCarlisle 'oJo

@HenriMenke bonus answer: because he uses a half baked format like manmac that doesn't have the refinements of a proper format like LaTeX where these things are taken care of automatically.

@HenriMenke @egreg fails to follow the instructions on where to answer:-)

@DavidCarlisle did you see the etex question?

@UlrikeFischer yes just looking, I think the kernel code is wrong...

@UlrikeFischer you have to love comments like % \changes{v1.0y}{1997/10/10}% {\cs{reserved@c} not \cs{verbatim@out} to save a csname} saving one entry in the string pool, now look how many entries @JosephWright has blown by \input expl3 :-)

@DavidCarlisle ;-).

@DavidCarlisle it is this redefinition of \ch@ck. Didn't we just had this?

Got the final results, thank you: twitter.com/paulocereda/status/1203989046374096896

@DavidCarlisle tex.stackexchange.com/a/515691/2388 and github.com/davidcarlisle/dpctex/issues/23

@UlrikeFischer yes sure (and I see you found a duplicate) but what I noticed just now is that the format filecontents does \chardef\reserved@c15 % \ch@ck7\reserved@c\write% so assumes 15 rather that using the full \e@alloc check which is aware that luatex has more writes

@DavidCarlisle ah. But would a correction of this then avoid the problem here? I thought it is due to the \if@filesw creeping into the \ifnumcheck.

@UlrikeFischer I haven't tested, but it might if it didn't use \ch@ck at all then it might not break if etex.sty changes that...

@DavidCarlisle but there are more uses of \ch@ck in the kernel e.g. in \newinsert, and etex takes care to change the definition here. Wouldn't it be its job not to break the kernel and follow updates if needed? (Unless the \ch@ck is actually wrong in the filecontents definition).

@UlrikeFischer well as I say it is wrong for luatex (the newinsert ones are OK as inserts have to be classical registers)

@DavidCarlisle so who will open the issue ?

@UlrikeFischer it's tricky as the job of etex.sty is precisely to break the kernel and re-instate the old code, so i suppose the consistent thing to do would be for it to re-instate the original filecontents code with a pre-2015 compatible definition, but I'm not sure that's what the user would want to be the outcome.

@DavidCarlisle but as the filecontents package is basically gone they would loose the overwrite option. couldn't etex simply do nothing if it detect an new kernel? Apart from the mt2pro problem which seem to have gone stale again I don't know any case which actually needs it.

@UlrikeFischer I had that for a while but it caused complaints. if a user is loading etex.sty to get compatibility with code that uses it \globcount etc, then it's reasonable to expect that after loading a compatibility package that they are defined.

@UlrikeFischer of course most people were not using any of the etex.sty commands and just loaded it to get more registers, so simply not loading it is the thing to do, but that is the easy case....

ooh

@DavidCarlisle we could create a etex-legacy.sty with the old code and people who really want it can load it instead ... (the normal etex could issue a message for it).

@UlrikeFischer or we could put a \relax after \ch@ck7\reserved@c\write% so #4 isn't \if@filesw.....

@DavidCarlisle would be easier, but you couldn't create a new repo ;-)

@UlrikeFischer I could fork latexe/latex2e

@DavidCarlisle ooh

@DavidCarlisle and then split it?

@UlrikeFischer you are mean

@UlrikeFischer github.com/latex3/latex2e/issues/238

@cis Don't think so. But if you can provide a typesetting-related use-case, you may convince Joseph to implement it for you ;-)

@AlexG OK.

@AlexG Or maybe, I should use SageTeX

@HenriMenke well it starts at the euklid tutorial. And looking closely at the text: In the "The Complete Code" there are random factors. There are coordinates using "rand". Could they slightly change the size of the pictures?

@HenriMenke I think this is it. I extracted on the images and measured its height and get values between 195pt and 206pt. And 11pt can certainly change the page break.

@DavidCarlisle did we change something for \@ifpackageloaded? In the last days I got constantly errors with \tableofcontents and I traced it down to a hyperref redefinition of \@starttoc (if the aux-file is empty). But I don't see what actually changed.

\documentclass{article}
\begin{filecontents}[overwrite]{\jobname.aux}
%%ensure empty aux-file
%\global\let\hyper@last\relax
\end{filecontents}
\usepackage{hyperref}
\begin{document}
\makeatletter\show\@starttoc
%contains \@ifpackageloaded {parskip}{yes}{no}
\tableofcontents
\section{abc}
\end{document}

@UlrikeFischer in git or in texlive?

@cis it seems you are installing something that is installed?

@DavidCarlisle Ah. I see. You replaced \ltx@ifpackageloaded by \@ifpackageloaded, and the first is allowed in the document.

@DavidCarlisle But I do not have SageTeX (in TeXLive 2019): "sagetex.sty not found"

@cis see tex.stackexchange.com/questions/519730/…

@UlrikeFischer grmbl I already backed out (or modified) some of those. I'll catch that as well although I started a message to the team list about that one, I don't think I sent it...

@UlrikeFischer OK

@cis :)

@PauloCereda Can you translate the phrase?

@cis no idea

@UlrikeFischer font setup is \@onlypreamble as it needs to be done before the fonttables are notmalised in \begin{document}, but \@ifpackageloaded, is \@onlypreamble to reclaim half a dozen bytes....

@DavidCarlisle I don't remember a message about this, but I couldn't swear that there was something - I was a bit distracted writing a document ...

@DavidCarlisle yes it took my quite by surprise that it can't be used.

@PauloCereda German "Schnauze, Sülze!" is nearly impossible to translate into english, it could be something like "Shut up, (you) baloney!"

@UlrikeFischer normally you can just save the definition with \let\zzz\@ifwhatever and then use \zz after the original has been disabled but for ifpackageloaded \@ifl@aded is also disabled so you have to redo the entire stack with names that are not going to be removed:(

@cis oh

@UlrikeFischer in hyperref.dtx I see \ltx@ifpackageloaded{parskip}{\parskip\z@}{}% ie it's still using the ltx@ one?

@DavidCarlisle perhaps you reverted it? I will reinstall hyperref from the git.

@UlrikeFischer no it is wrong on ltxcmds branch, I was looking at master

@DavidCarlisle yes just saw it too.

@UlrikeFischer I'll fix later (Hy@ifpackageloaded I guess

@DavidCarlisle why do you have to redo a stack? Isn't \@ifl@aded simply testing if the ver@package.sty csname exists?

@UlrikeFischer I do not know, what to do now. pythontex sagetex.ins ? No...

@UlrikeFischer I mean I can't simply do \let\zzifpackagloaded=\@ifpackageloaded then use \zzifpackagloaded as you'd then get an error on \@ifl@aded as that is separately made an error after begin document

@DavidCarlisle ooh zz

@PauloCereda I choose macro names that @UlrikeFischer can't type, as she is mean.

@DavidCarlisle ooh

@DavidCarlisle you could create a repo for each zzz-command - I could never get write access.

@DavidCarlisle ooh she is mean

@UlrikeFischer ooh a plan

@UlrikeFischer try now....

@PauloCereda Hey, we need a rule sagetex.yaml

In later times, I could try to write that, but today....

@DavidCarlisle works fine now.

@cis well assuming that my guess that you have windows was correct: either change the PATHEXT variable so that it no longer contains .py, or change in sagetex.ins all .py extension to eg. .pyxy, extract and then rename the created files back.

@UlrikeFischer mail to team sent....

@UlrikeFischer This seems to work. Now I need to finde out, how to sage a file with path....

sage: /home/sage/Desktop/Test/SageTeX/sagebsp01.sagetex.sage
  File "<ipython-input-2-89a2f91aed61>", line 1
    home(/sage/Desktop/Test/SageTeX/sagebsp01.sagetex.sage)
         ^
SyntaxError: invalid syntax

@DavidCarlisle tex.stackexchange.com/questions/520670/…

@DavidCarlisle Yes: I think we should cut down \@onlypreamble to things that are conceptual rather than token-saving

@JosephWright er coincidence?

@JosephWright irony is it was only there to save tokens and then a stack of packages just duplicate the entire definition to avoid \@onlypreamble which doubles token memory and csname use:-)

@JosephWright my example was better ;-)

@DavidCarlisle @)

....

@cis never trust software written by a duck wielding a cricket bat

@cis I have asked you before not to refer to me as Mr Carlisle

@cis it's the wrong title and even had it been the right one it's impolite to insist on referring to someone using a form of address they have asked you not to use.

@cis I'm not German. (but if you used the correct title it makes no difference as I would ask you not to use that form either. I prefer not to use a title on the web)

@cis -- "Madam" would be out of place in more ways than one.

@barbarabeeton you don't see me as a Madam?

@barbarabeeton Whats about "mistress"?

@DavidCarlisle -- Nor as "Madame". I don't think "mistress" has quite the correct implication either; as far as I know, you're gainfully employed, not kept.

@barbarabeeton Also, likely not a witch

@UlrikeFischer Okay, I'm now re-seeding the RNG before every use. I hope that fixes it and I hope I caught all of them.

@HenriMenke @UlrikeFischer @JosephWright splitting off any dependencies of hyperref, plus a couple that have reported issues fixed: auxhook bigintcalc etexcmds gettitlestring hycolor intcalc kvdefinekeys kvsetkeys ltxcmds refcount soulutf8 uniquecounter oberdiek

@DavidCarlisle Is there a timeframe by which the oberdiek split will be finished? I don't want to release the new PGF before that.

@HenriMenke that is the last split that I plan, although as I say, if there are reported issues that get fixed I may split off individual packages, but unlikely to be before the new year now.