« first day (5122 days earlier)      last day (28 days later) » 

00:17
@DavidCarlisle -- Please don't say that too loudly. There's someone in this country who might overhear it. I'm not anti-monarchy, but much prefer that the monarch be someone with a brain and a sense of decency.
cfr
cfr
01:12
how do I stop l3build check searching installed TL? checksearch = false is apparently not enough. do I need to do something else?
which makes me think setting TEXINPUTS isn't doing what I think either ....
 
6 hours later…
06:45
@cfr That should be enough - what are you seeing?
@cfr We only suppress the TeX and Lua trees at the moment - if it's something in another tree, we might need to extend
cfr
cfr
07:25
@JosephWright it picks up things in the fonts tree:
/usr/local/texlive/2024/texmf-dist/fonts/enc/dvips/base/8r.enc
/usr/local/texlive/2024/texmf-dist/fonts/enc/dvips/cm-super/cm-super-ts1.enc
/usr/local/texlive/2024/texmf-dist/fonts/enc/dvips/lm/lm-mathit.enc
/usr/local/texlive/2024/texmf-dist/fonts/type1/public/cm-super/sfrm1000.pfb
/usr/local/texlive/2024/texmf-var/fonts/map/pdftex/updmap/pdftex.map
@JosephWright and the build is also not isolated, I guess, because i copied your strategy. (I should have read the .cnf file again first.)
07:45
@cfr We can add the font tree to things that are isolated - please log an issue
@cfr Not sure the map file will be so easy, but the fonts in dist should be doable
08:27
@cfr Fonts look a bit more tricky that TeX/Lua inputs - there are a lot of variables here!
cfr
cfr
08:38
@JosephWright yes, that was my initial reaction, too. pdftex.map is ok if found. as log as it only picks up the map files updmap generates, that's easy to deal with. but .enc files are really easy to screw up. at least, my notes tell me I've redone ctan uploads for enc-related errors rather a lot ....
@cfr :)
@cfr I think this might be one to ask on the TL list - surely there's some simple way to skip all of texmf-dist/fonts in the same way there is for other stuff - or perhaps there's a way to skip all of the installed tree that we missed before
cfr
cfr
@JosephWright yes, probably. let me think about what I want it to do first ....
@cfr Cool
@cfr Once we know what to ask, lets see if something obvious is being missed
09:25
@JosephWright @cfr one way is to move the binary so then it's auto-locate will assume texmf-dist root based on the parent directory of wherever you moved the binary to, but then you basically are responsible for setting up everything
 
3 hours later…
12:35
I'm installing Win11 and am wondering about sharing my TL installation between macOS and Windows- anyone know how to easily add 'foreign' binaries to an existing install?
12:52
@JosephWright just install one then from there tlmgr install the binaries other (with a tlmgr package name i forget) then set path on other machine to that it worked well when i had that set up but i don't currently
@DavidCarlisle Yeah, I was hoping there was some simple way to do the ' tlmgr install the binaries other' - I'm not keen on having to work out every separate one to install
@JosephWright Are you looking for tlmgr platform add ...?
@samcarter That sounds like it, yes :)
@JosephWright I can see all the packages like xmltex.x86_64-cygwin but there was a collection somewhere that does them all (or just do a minimal install on the second machine and just copy the bin directory would work in practice)
@samcarter I'll do that once I have my Win11 system set up - currently I'm copying all of the TL installs off my Win10 VM to a backup disk - around 100Gb so going to take a while
12:59
@JosephWright ... an impressive collection! :P
@samcarter TL'09 onward - everything you can install on Windows, basically - macOS makes it tricky to install older systems, hence the query
I see Javier has put my news in his news :)
@JosephWright yes, I lost all < 2019 when switching to M1 :(
He is a member of the team ...
@samcarter Indeed - I'm still pondering whether to stick with a Mac next time, or switch back to Windows to keep x86
@JosephWright you should really deduplicate your TL folders and use links
@Skillmon I likely will - put the files on the host OS and make a share to the VMs
So something like T:\2024 - a nice drive letter for TeX Live :)
13:01
@JosephWright You can also deduplicate your individual TL versions. 80% of TL2010 is TL2009.
(80% was a random number, not scientifically correct or verified)
@Skillmon That's more complex, and I think safer to keep separate
^^^ This gave me around 20 GB back :)
@samcarter there is another script with a different approach around, I'd have to dig for it, which I can't on my dayjob PC.
Not sure which approach is better though.
The other one is build atop hashes of files and creates a hidden folder containing all the files with their names replaced by the hashes which then get hardlinked to the files in the original folder. That way all duplicates get replaced (but you could lose files if there's a hash collision)
@samcarter I'm not so fussed about space per se, more about the effort of installation and keeping things up to date
@Skillmon TL'09 is about 2.77 GB, TL'24 about 8.95 Gb :)
@JosephWright well, that only concerns the latest TL, for all older versions you can easily save a bit of space at the onetime cost of a few minutes processor time and read cycles on your HDD.
@JosephWright yes, I know, hence I put 2009 and 2010 next to each other :)
13:09
@JosephWright ... all the automatic backups of easybook take space :)
@samcarter :)
@Skillmon I like being able to 'just look' at older TL's - and I want something that works on FAT32 ;)
13:29
@JosephWright your fault for using the worst file system that's still around :P
@Skillmon My VM is NTFS, but memory sticks are usually FAT as it's usable anywhere
13:51
@JosephWright Well, NTFS supports links. If you then copy to FAT you'll get those links dereferenced and the correct files.
@Skillmon I'll think about it once I have things copied across
 
2 hours later…
15:52
@DavidCarlisle Hm, no clue what is happening. I cannot reproduce with non-tweaked fonts either...
16:08
@DavidCarlisle the luatex manual is rather underdocumenting this command ;-(
@DavidCarlisle oh you commented in the issue, I was a bit distracted by my nemesis cleveref
@mickep we could simply blame you and close the issue?
@UlrikeFischer @mickep I'll try and make a plain tex example later
16:54
@UlrikeFischer @mickep axis works with stix2 here
\input{luaotfload.sty}

\font \stixm=STIXTwoMath-Regular.otf
\textfont0=\stixm
\textfont1=\stixm
\textfont2=\stixm
\textfont3=\stixm
\let\tenrm\stixm

$$ a+ \Uvextensible height 30pt \Udelimiter 5 0 `| $$
$$ a+ \Uvextensible height 30pt axis \Udelimiter 5 0 `| $$

\bye
@DavidCarlisle has the | a size of 30pt for you? Here it doesn't extend.
@DavidCarlisle try height 30pt depth 30pt, for example
@UlrikeFischer hmm no but axis works:-)
Here the top one has height .5cm, the second height .5cm depth .5cm axis, and the third one as the second but without axis.
@mickep so what is the magic Uvextensible syntax (without the context wrapper, or even with)
17:06
@DavidCarlisle I only used \Uvextensible height .5cm depth .5cm axis \vert for example.
@mickep hmmm thanks
@mickep the manual is not exactly clear but is that the same as \Uvextensible height 1cm axis \vert as axis shifts it anyway (that seems to be what latex is trying to do)
@DavidCarlisle actually I doubt that. If you look at the Uleft/Uright examples it looks as if axis simply shifts the line a bit up.
@DavidCarlisle no, height 1cm axis uses no depth and shifts up. See upcoming screenshot, where height .5cm depth .5cm axis is to the left and height 1cm axis to the right.
@UlrikeFischer oh well in that case that would explain the latex output if it means we are setting height above the math axis to teh big size and depth below the math axis to 0
17:13
@mickep thanks then I think we are getting bad output due to bad input. shocking.
@DavidCarlisle yes, but it doesn't explain yet, why some fonts seem to work anyway as the issue claims. But I need to test that.
@DavidCarlisle Yes, that is indeed shocking.
@UlrikeFischer I haven't looked at the glyphs in those fonts, or the various sizes it has. That could explain it partly at least. (some fonts have glyphs sitting in weird places, and I have a vague memory that the bar is one of those glyphs)
@UlrikeFischer StixTwo has no variants but go directly to extensible.
@DavidCarlisle looks as if some fonts gets it right by accident. So we should change the definition. The main question is where and under what conditions (curiously it seems to work even with legacy math?).
17:29
@UlrikeFischer the manual has some words about this working for tfm:
LuaTEX internally uses a structure that supports OpenType ‘MathVariants’ as well as tfm ‘exten-
sible recipes’. In most cases where font metrics are involved we have a different code path for
traditional fonts end OpenType fonts.
@UlrikeFischer I'm good at reading manuals, as you know:-)
@DavidCarlisle I trained you well ...
@UlrikeFischer not really, just threatened me with a sword if I did wrong.
A package with "wheel" in its name solves problems with triangles: tex.stackexchange.com/a/730264/36296
@samcarter that question desrves a picture mode answer
@DavidCarlisle we are waiting for 11 years for you to post one. Nobody else dared!
 
1 hour later…
cfr
cfr
18:44
@JosephWright what is the status of l3build functions which don't specify usage? I'm don't know if I need to duplicate them or not ...
18:58
@DavidCarlisle Somebody should save that...
Good that it got solved(?)
19:21
@cfr Which ones?
cfr
cfr
@JosephWright right now, localtexmf(), but probably others. and dep_install() has private with a question mark.
@cfr I'm sure we can make them public with a reasonable use case :)
cfr
cfr
@JosephWright define 'reasonable' :)
@cfr 'Someone needs it'
cfr
cfr
19:40
@JosephWright I'm trying to mirror what check does so in fontscripts so I have isolation that (1) actually works and (2) stands some chance of building somewhere other than my machine. (well, berenisadf is not going to build in most places because it has external dependencies, but I'd like it to at least build on github or in a chroot without I have to cheat.)
cfr
cfr
19:57
does anybody know why I lose most of the formatting if I attempt to edit this answer:
19
A: How to install "vanilla" TeXLive on Debian or Ubuntu?

cfr PLEASE THINK BEFORE EDITING! Please do NOT edit this answer in ways which render it distro-specific. Although this question specifically addresses Debian/Ubuntu systems, it serves to provide information to users of GNU/Linux more generally. The way things work on your system may not be the way...

even if I make no changes the preview shows no formatting after a little bit at the top.
@cfr I see same if I hit the edit, I think it's because of a change in the formatting of code blocks <!-- language:bash --!> what you see is the cached rendered html page but as soon as you hit edit, it regenerates html using the new interpretation which doesn't understand it
cfr
cfr
@DavidCarlisle ah, thanks. is there an equivalent or do I have to mark the language for every block?
@cfr not sure, I'd just use ```bash I think
cfr
cfr
20:17
@DavidCarlisle se removes ``` in pasted text?!
cfr
cfr
@DavidCarlisle I edited the answer elsewhere, but I can't copy it back without something messing it up. I assume se, but maybe something else.
@cfr just looked now and you have edited it and it looks ok?
cfr
cfr
@DavidCarlisle yes, because I replaced the deleted back ticks by hand.
Daily TikZ show off. 😁😁😁
20:37
@Atcold Why different colors for the subscripts in y_0, y_1 and y_2? Also, the plot has a bit too much "clutter". I suggest that you make it simpler.
@mickep ouch! You're correct. Those subscripts should have actually inherited the colour (I do differentiate between coloured subscripts and non-coloured ones, and this case wanted the coloured ones. Good catch!).
@mickep about the “clutter”, all my drawings are densely annotated. I like them to be extremely descriptive and informative.
@Atcold Sorry if it came out too negatively, the figure is nice. I would remove all the numbers, but it is a matter of taste.
@mickep haha, yes, I could do that. This is not a “cartoon”, though, but an actual diagram with data points. The 3d version of the base diagram that follows, which has the correct axis labels.
@Atcold The thing is, it is difficult to see where all those 1s and -1s actually point. But if you need them, you need them, I guess.
20:54
I'm going to release my book soon, on GitHub. It would be amazing having LaTeX feedback from detail oriented people!
Is it frowned upon to do \use:c{l_mymodule_#1_tl} instead of \tl_if_exist:cT{l_mymodule_#1_tl}{\tl_use:c{l_mymodule_#1_tl}} in a place I don't know l_mymodule_#1_tl exists? I know this is just the behavior of \csname...\endcsname but I want to know if it's bad practice
@mbert if you need to ask, it's probably a bad idea
@DavidCarlisle Hmm but I thought it was supported behavior that \csname foo\endcsname did nothing if \foo doesn't exist
21:10
@mbert well it is is, but when you get some email in 30 years time from someone expecting you to remember how to support your own code, which version is more obviously a conditional branch on the existence of a module...
@DavidCarlisle Well by then everyone will be using context, so... :)
@mbert I stuck out afterpage as some quick comp.text.tex answer and thought by now everyone would be using latex3, life doesn't always go to plan
22:18
I've recently begun to understand how the \foreach command in the tikz package works through the tikz manual. It does seem to require some creative thinking for practical applications, however, and I was wondering if anyone has or knows examples where the \foreach loop was very advantageous to use?
@Atex hard to know what you are asking \foreach is a loop (it is a bit weird as it places each iteration in a group) but if you need a loop you need a loop, so it is not a question of advantages and disadvantages
@Atex there are over 20 thousand posts with examples on this site: tex.stackexchange.com/search?q=%5Cforeach
I mean the practical applications of it. I'm very well aware that it's essentially just a tool for saving time instead of repeating a command, but you have to sometimes write it in quite a different way when "translating into a loop", so to speak. I think visual demonstrations would help me to recognize when it could be beneficial for me to use it
@Atex as I say there are tens of thousands of examples here, in most cases if you need a loop there isn't a practical alternative of not using one. just think of drawing some graph paper where you want to draw a rule at some regular interval, in tikz as in any other language you want to say for x in 1..n draw a line at x what else would you do?
22:33
yeah, I guess I'll just have to look through the questions to see if something inspirational catches my eye. I just thought maybe there were practical demonstrations for these things somewhere other than in the tikz manual where people had used them for real purposes.
Right now, the only situations I can think of it being useful are what the tikz manual covered. So for making a clock, coloring balls, changing colors as you gradually draw more nodes etc. Very limited, but I'll look through tex and see if something creative appears.
@Atcold "detail oriented people" ought to be the most euphemistic version of "nerds with typography OCD" I've ever heard of.
3
@Atex I'd have thought most largish examples use it somewhere, i just took a random first example at tikz.net and as expected that one example used\foreach. Whether or not you consider it practical I have no idea: it was just a demonstration that 100% of the samples I tried had that command tikz.net/…
22:51
@DavidCarlisle I see. It appears that what I was concerned about is true: Most practical utilizations of the \foreach loop requires using it multiple times... makes sense since you can only create 1 "repeated pattern" for each loop. It also adds another layer to the amount of abstract thinking to apply it, however. Will be a difficult command for someone like me to use in that case for the time being.
23:36
@Skillmon the Romans thank you
@DavidCarlisle <3

« first day (5122 days earlier)      last day (28 days later) »