« first day (5227 days earlier)    last day (1 day later) » 

cfr
cfr
03:41
@DavidCarlisle you don't have to make a virtual font to change the kerning for 8-bit. you can just adjust in the tfm. gentium-tug included a whole bunch of changes to metrics as part of the conversion from ttf, but you could obviously do the same starting with type1. I assume you have to go virtual with luatex, though, since you're just using the fonts.
@MaestroGlanz if you mostly want to change kerning, you could also look at what microtype can do.
@DavidCarlisle yes, but it does that because microtype affects maths differently with pdftex but not luatex.
@MaestroGlanz it is easier to use arbitrary fonts with luatex. it is easier to compensate for the deficiencies in a font with traditional tex fonts (inc. type1), at least if you want to package the result. though it depends on the deficiencies, of course. as far as I know opentype and truetype are not the same. opentype come in postscript or truetype 'flavours' and these correspond to different ways of drawing the glyphs, if I've not forgotten completely.
@MaestroGlanz so what kind of fonts do you actually have? and do you want to use pdftex or luatex?
 
3 hours later…
06:54
Several. It is mainly the Monte Carlo by Rob Leuschke and the Slavkappen. The Slavkappen is in my opinion a very good font, but the long s is just wrong. From what I got so far.
1. It is better to edit the font than to fix it on LaTeX-level.
2. The fd-file tells pdfLaTeX, what to use from the font. It is not necessary by LuaTeX, but usable by it
3. For Monte Carlo it makes only sense create an fd-file to make the private area in the utf coding accessible. But in doubt, better stick to Lua(La)TeX.
You can see here, what I did with that:
https://bitbucket.org/maestro-glanz/liederbuch-sondergut/raw/30f05e6e2f80fba975b8fff6fc197ec4c7ecd6c5/Erlangenlied_Wandbild/Erlangenlied_greg_drei_Strophen.pdf
It's a gregorio with Slavkappen.
cfr
cfr
07:42
@MaestroGlanz you can't make the PUA available via an .fd file. it doesn't provide that kind of information. I don't agree with (1). if it is your font, of course that is different. if it is not, but you have permission, it may or may not be better. gentium-tug could have edited the fonts, since they were OFL. there's a reason the changes were done on the TeX side rather than the font side. (maintainability, basically.)
@MaestroGlanz re (2) yes. LuaTeX can use .fd files. but you wouldn't usually use them with luatex except for 8-bit fonts, if you needed that. you wouldn't use them with unicode fonts. I suppose the exception would be if you wanted to access a truetype or something set up for pdftex in just the same way for some reason, but hard to see why.
@MaestroGlanz (3) no. as above, you can't make the PUA available by adding an .fd file. look again at the examples @DavidCarlisle and I posted. fd files map NFSS font characteristics to (some kind of) font.
@MaestroGlanz are you trying to do this for opentype, truetype or type1 fonts? are you trying to do it for pdftex or luatex? right now, you are just making a muddled confusion, I think, because there are too many variables.
@MaestroGlanz nice ;). so you are using luatex and can forget all about fd files, tex font metrics and the rest. :-)
why does github keep complaining that my main branch isn't protected? I followed the information link and it told me a bunch of stuff about something-complicated-sounding which it said is entirely different from protection, but nothing about the (hopefully)-less-complicated-thing the complicated-sounding-thing is much better than.
08:41
@MaestroGlanz as @cfr says (1) only when you have a licence to do so (most tex font packages are not maintained by the designer of the underlying font) (2) yes (3) no. As said before, fd files map to file names, they are unrelated to any internal font data such as private use code points.
08:55
I use only OFL fonts.
@cfr I create confusion, because I lack clarity myself yet.
Because github is not what it used to be. (Very well and nicely phrased)

I think, I will move to gitlab. Github is bothering me with 2FA. Bitbucket has no issue section. You have to set up Jira-SM for that. No thanks....
@cfr Yes, but I had to hack the kerning. I prefer solid solutions over dirty hacks.
@cfr You can just turn that off - it's trying to be helpful but for small projects it's not useful
@MaestroGlanz Well 2FA is a good idea, so ...
I disagree... If someone has access to my machine, he will inject code. If he hasn't he has to get access via github. And this means, that they hadn't proper brute force protection. And this is not difficult. There aren't a lot of scenarios, where an attacker has no access to my machine and benefits from no-2FA. I agreee for other projects. If I maintain the source code for the SSL-encryption, then it makes sense.
cfr
cfr
@JosephWright how do I do that?
@JosephWright But I have to set it up once. And putting my phone number into Github creates more threats than it eliminates.
@cfr Er, now that I don't remember - you only do it once per repo ...
@MaestroGlanz 2FA doesn't have to mean phone-number-based (it's well-established that SMS 2FA is the weakest approach of the available 2FA systems)
@MaestroGlanz I only mean that I see why they are pushing it - in the end it is of course your decision what services to use
cfr
cfr
09:06
@JosephWright you can give your phone number directly to ms instead :-)
(For personal use - my work for example insists on 2FA for accessing university data off-site)
cfr
cfr
@JosephWright I get asked even on site. only how do you use 2FA when your 2F is the phone you're trying to use in class?
so, in effect, I can't access e-books while teaching ...
(at least, not if I want slides up at the same time.)
@cfr We need 2FA for SITS/eVision access in all cases, otherwise for the VPN or off-site login - we've not got 2FA for logging into teaching machines in lecture theatres (partly as phone signal is poor on campus)
@cfr I have colleagues who have Yubikeys - they therefore can't do anything ona phone at all :)
@cfr Thanks, I did not know pdftex supported expansion for math.
Off-topic question: Who of you uses Linux/Mac/Windows
09:11
@cfr I have stuff logged in on my phone - I just have to switch to my authenticator (Authy) as required]
@MaestroGlanz Mac with a Windows VM for some work stuff
cfr
cfr
@JosephWright we actually don't need it for vpn. (or didn't - I haven't needed it is a while.) it will let me access the vle, but i can't look at any e-books, which makes no sense. it it was confidential info or something, I'd understand, but they expect people to do registers on phones ....
@MaestroGlanz linux. win in class.
@MaestroGlanz Oh, yes, like @cfr says - work teaching systems Windows
cfr
cfr
@JosephWright this I can't figure out. if I switch to the authenticator, when I go back to the login page, it reloads and the code is no longer valid ...
@cfr Ah, yes - that can happen
cfr
cfr
@JosephWright so I can see student details and contacts without 2FA, but I can't read the meditations in class ...
09:15
@cfr :)
cfr
cfr
@JosephWright well, obviously I actually have paper. but I was trying to help a student by showing them how easy it was to find the right text through the reading list ...
cfr
cfr
@JosephWright is there a way of telling l3build that something in typesetdeps needs an additional step to install? there doesn't seem a nice way of triggering updmap.
@cfr Not at the moment, no
@cfr Wouldn't that be an issue? I thought one should only ever run updmap from the admin account
cfr
cfr
@MaestroGlanz I know. but I think you need to decide: type1/pdftex or opentype/luatex. because you need quite different approaches and I don't think trying to figure out everything at the same time is helpful.
@JosephWright well, I was thinking github. but it would be ok if you used files under build for output.
@JosephWright it doesn't actually matter the account, as I'm sure you know. one should only ever run updmap systemwide unless one knows what one is doing.
09:35
@cfr Then I better create a package, that is just an interface for LuaTeX to make the alternates and everything more accessible.
[ 9/13, 12:25/02:07:59] update: pst-geo [157594k] (60387 -> 74167) ... ^C
Err ...
cfr
cfr
10:04
@JosephWright this?
..... pst-geo.tex
0.09  2025-02-20  - added different increment for x and y coordinate
@cfr Must be!
cfr
cfr
@JosephWright ctan seems confused. it still lists the version as 0.07? but the top level files are new ...
@cfr Oops
cfr
cfr
@JosephWright apparently nobody has noticed since 2021 when 0.08 was released and the data directory list has all 2016 ... seems a bit weird?
@cfr Yeah - report to them?
cfr
cfr
10:12
@JosephWright later, yes ...
10:35
@cfr TIA
11:10
has anyone else also problems with texdoc? It no longer opens anything for me.
ah back to normal. I had to kill an adobe process.
@DavidCarlisle youtube is adding text and in some parts you get rather confusing stuff ;-)
Hmm yes I can probably disable captions in the youtube interface somewhere, or add my own, or I could simply blame you
@DavidCarlisle oh it is interesting, also the various ways to read align ;-)
@DavidCarlisle oh, the law tex team. I will now be sued.
@UlrikeFischer English spelling and pronunciation is so regular
12:05
@mickep ask @UlrikeFischer to supply a /Phoneme tag
@DavidCarlisle lejtäch
 
5 hours later…
cfr
cfr
17:09
@JosephWright ??
@cfr 'Thanks In Advance'
 
1 hour later…
18:34
@cfr I would say: either use expansion zfor all glyphs in math or not at all. I guess that is what happens in luatex since all math glyphs cone from the math font. In my opinion what happens in pdftex there is on the border to be a bug. Not sure where, though. @DavidCarlisle?
@mickep as you say it depends a bit on what you expect \mathrm to do if it's using the "same font as text" in a context where the text font is being transformed, I (think) latex could ensure the mathrm font was not subject to microtype without any changes in pdftex
@DavidCarlisle I think it would be good to do. It is a bit arbitrary if plus gets it but not minus. For example.
cfr
cfr
19:27
@mickep no, it doesn't depend on unicode-math. there's no expansion in luatex even with the default maths config, whereas there is with pdftex.
@DavidCarlisle but it isn't just \mathrm. e.g. $a + b$ gets protrusion and expansion in pdftex, whereas it only gets protrusion in luatex. (which is what I was trying to say re. the muskip business the other day, albeit unsuccessfully.)
@DavidCarlisle I'm not sure that would be a good solution since microtype apparently offers application to maths as a feature?
19:54
@cfr Oh, OK.
@DavidCarlisle Btw, is there a list of examples somewhere of situations when italic corrections is added when it should not be added and not added when it should be added?
20:37
@mickep not that I know of (@UlrikeFischer may have something) but we're a bit distracted with a new -dev release and tagging issues just at the moment:-)
@DavidCarlisle Oh, well, good luck with the tagging...
21:24
@mickep you only need to search a bit on the main site, the topic came more than once, e.g. tex.stackexchange.com/questions/734370/… or tex.stackexchange.com/a/253531/2388
22:09
@UlrikeFischer Thanks for the links. I am not sure I will catch all cases. (I'm not sure about the period case, if that really applies...)
@cfr Apropos our discussion of what's a word, a colleague at MIT who does a lot of language revitalization work with Native American languages just posted the following:
Passamaquoddy-Wolastoqey word for the day: yalanokiye 'he/she walks around on the ice in order to enjoy the sound of it cracking'
cfr
cfr
22:38
@AlanMunn I wrote a question, but I fear it will offend you.
22:53
@cfr I'm not easily offended. :)
23:06
@AlanMunn we could take that as a challenge
@DavidCarlisle You are mean.
Jun 29, 2017 at 16:15, by Paulo Cereda
@DavidCarlisle you are not mean :)

« first day (5227 days earlier)    last day (1 day later) »