@samcarter why do none of the answers suggest the obviously superior method used here?

@DavidCarlisle ooh

@DavidCarlisle Seems everybody is happy with picture mode :)

@samcarter ooh twitter.com/MartinaGluscevi/status/1574560092749430790

@DavidCarlisle Ooh, you are a true artist. :)

@PauloCereda oh, very efficient!

@UlrikeFischer I thought about making the message "guard against @UlrikeFischer" github.com/davidcarlisle/graphics-pln/commit/…

@DavidCarlisle ;-)

@UlrikeFischer I was tempted not to restore the outer version but I suppose it is better this way

@DavidCarlisle why? does anyone care if it is non-outer?

@UlrikeFischer some ifx tests may fail or @barbarabeeton someone may actually like outer and want the error check there.

I'll push to ctan as it is, I think.

@UlrikeFischer mkctan script is so much quicker than l3build

$ time ./mkctan
  adding: graphics-pln/autopict.sty (deflated 74%)
  adding: graphics-pln/exmplcol.tex (deflated 11%)
  adding: graphics-pln/exmplpfg.tex (deflated 49%)
  adding: graphics-pln/graphicx.tex (deflated 58%)
  adding: graphics-pln/picture.tex (deflated 58%)
  adding: graphics-pln/README.md (deflated 60%)
  adding: graphics-pln/color.tex (deflated 56%)
  adding: graphics-pln/exmplgrf.tex (deflated 30%)
  adding: graphics-pln/exmplpic.tex (deflated 34%)
  adding: graphics-pln/miniltx.tex (deflated 70%)

No tests, no documentation, no Lua:-)

@DavidCarlisle but we should do something about the "no eplain tests" in graphics-def ;-)

@UlrikeFischer yes I'll push now then once it is in tl we can add a test in graphics-def that passes

@DavidCarlisle we could at the same time separate the engine tests and no longer test the luatex driver with xetex.

@UlrikeFischer yes

@UlrikeFischer unicode-math doing everything at begin document almost as annoying as \outer

@DavidCarlisle ;-).

@DavidCarlisle It's one we really could to to integrate into the format, if we could work out how to do it

@samcarter circledsteps.sty is picture mode... ;-) (well, pict2e, but I think it qualifies)

Am I the only one watching the current chess drama?

@yo' I've been keeping an eye on it

@JosephWright cool :)

@yo' I am watching it too!

@Rmano :P

@samcarter First, if the behavior exposed in the question is NOT reproducible, the answers, technically, solve a non-problem. They might or might not be useful for other (or similar) reasons, but NOT for the reason of a no-longer existent behavior. The answers we have are simply in the wrong place these days.

Second, to me, it does matter whether the numbers touch/intersect the circles or not. Most folks can live with a slightly off-center circle around a number as long as the circle is somehow around the number and does not touch it.

Ooh, an alternative Bonum math alpha in the making. :)

@GeekestGeek well do this numbers touch or not:

@UlrikeFischer At your resolution, the digits do touch the enclosing circles. What's your point?

@GeekestGeek "The answers we have are simply in the wrong place these days." This might or might not be the case, but aren't they better left there instead of closing the question? I'm glad the close review unanimously voted to leave open.

@GeekestGeek what do you know about the resolution of the screenshot from 10 years ago? Or about the resolution of other people than you which perhaps care that this touching here is addressed too? Why don't you accept that the answers are useful even if there is no code in question and only a not so sharp screenshot.

@samcarter Technically, NOT the answers were flagged but the question. So, strictly speaking, you shouldn't ask about the answers (in particular, whether the answers are in the right or wrong place) but about the quality of the question itself.

@GeekestGeek sorry but you are wasting my time with your pettiness. If you want to improve the site handle the really bad questions.

@UlrikeFischer It's less about the answers than about the question. David said that the question has no code, and so we don't know how the image was produced. It's meaningless to compare apples and pears, or, in the particular case, a fruit of an unknown name with an apple.

@UlrikeFischer Typesetting IS about pettiness. If you don't wish to talk to me, it's your choice. Heads up.

@GeekestGeek I do know how the image was produced. I even know why it looks different today. But I don't care, I don't think it matters for the question to be useful and one could replace the screenshot even with handdrawn circles without loosing something.

@GeekestGeek I did say that but I didn't say it should be closed. I just observed that comparing with your current example is a inexact science (in particular you get different effects at different font sizes)

@UlrikeFischer Earlier, 20-30 years ago, you would have been right about handdrawn stuff. Today, we can do better.

@DavidCarlisle You said I could vote to close. Since I have too low a rank to vote to close… Of course, the mods cannot do anything about their question in their role of moderators (i.e., using the moderator tools), but, perhaps, in their role as users.

@GeekestGeek yes sure you (or someone) could vote. Personally I wouldn't, that's the point of votes, people can vote differently.

@UlrikeFischer Yes. Perhaps, the weakest point of the question posed in 2010 is not the resolution of the images but the missing source coude. The resolution would be only the second weakest point, and not the main one. Thinking about it, for that particular question, even handwritten images could work as a last resort, similar to what Ulrike said. Though, as a rule of thumb, I would strongly oppose to posting anything handwritten.

@UlrikeFischer Did you look at my suggested branch for media setting?

@JosephWright öh no. I missed that. I will do it after the LWG meeting.

@UlrikeFischer Cool

@JosephWright can one with l3build use support files from another test folder? I was just trying to split up the graphic-def tests and can't get it working without copying the support files in every test folder.

@UlrikeFischer I'd probably use a common support folder, separate from any of the test ones

@UlrikeFischer ah I just did this:-)

@DavidCarlisle oh did you already push it?

@UlrikeFischer yep

LWG ah now? @JosephWright

@DavidCarlisle Work meetings

@DavidCarlisle Start of term here: this is a mad week for me

@JosephWright no i meant I forgot until you mentioned it.

@JosephWright yes we were delivering "new intake" at the weekend...

@UlrikeFischer is Frank around?

@DavidCarlisle He's away

@JosephWright ooh

@JosephWright there is a typo, \cs_if_exist_p:N \IfDocumentMetadata should be \cs_if_exist_p:N \IfDocumentMetadataTF. Beside this the logic looks ok, I only wonder if the code or \@@_backend_pagesize_set:nn should check if the arguments are > 0pt in case there are classes which don't set \paperwidth and \paperheight.

@UlrikeFischer FYI I delayed the luaotfload release a bit since Hans send me some proposed fontloader changes for testing which should probably become part of the next version.

@MarcelKrüger ok, fine with me. Do you think we could also get something like the goodies with fixes for fonts?

@UlrikeFischer We probably could, but I don't like goodie files. The whole idea of fixing fonts on the fly is rather problematic when combined with e.g. HarfBuzz shaping where some fonts can't be patched at all and we already have custom font features which can be used for such fixes when necessary. Do you have some specific advantage you want to get from goodies?

@MarcelKrüger mainly the idea to have prepared solutions that users could load. Currently adjusting fonts seems to be something only for people with some lua skills (or asking the right question so that you answer), and if they develop something it is burried in some document and then gets lost again.

@UlrikeFischer I see, I'll think about good ways to implement something for this.

@MarcelKrüger perhaps the only thing that is needed is some documentation how to create, name and organize "font fixes".

@UlrikeFischer an alternative for free fonts at least could be to fork/fix the fonts, not sure how much font tooling that would require.

@DavidCarlisle That sounds like a good plan.

@mickep Glad you said that, as part (b) of the plan was to get you to do it.

@DavidCarlisle Ooh no

(We have been converting the rsfs fonts into opentype and also fixed (well, one could say it is not beautiful) the \alpha in Bonum in the last week, so ...)

I don't think I'm the right person to go over the gyre fonts in general, though. It feels overwhelming, and there are others who handles the tools much better than I do.

@mickep a list of potential fixes, and some herding of volunteers might be enough

@DavidCarlisle I think best would be if someone could convince (i.e. pay) the Polish guys who did the fonts. I know there has been some discussions, but I don't know what has come out of it.

@DavidCarlisle One problem is that the standard is/was simply not there, and that people have made different choices. What is right and what is wrong?

@mickep yes if they are willing

@DavidCarlisle Hans wrote this to the list today.

@UlrikeFischer Aren't \paperehight and \paperwidth required valid parameters (@DavidCarlisle?)

@JosephWright minimal.cls doesn't set them.

@JosephWright not really, they are set but not used by the format or standard classes

@UlrikeFischer OK, I'll add some defensive code

@UlrikeFischer OK, both changes made

@JosephWright did you already merge? (It wasn't easy to test as the branch didn't have the new object code ;-))

@UlrikeFischer No, not merging till we are happy

@UlrikeFischer Should I reabse onto main and force-push?

@JosephWright for me it looks fine.

@UlrikeFischer OK, I'll merge

@UlrikeFischer I need to sort the case issue too so will do those then can do a release

@UlrikeFischer Have the bigger case-change thing I'm working on, but that needs a bit more time yet

@JosephWright which reminds me: I had to reinsert a test for color models \cs_if_exist:cT { __color_parse_model_ #2 :w } in the colorspace patches because it broke an example in TLC3.

@UlrikeFischer Can you send me an example?

@JosephWright I can try.

@UlrikeFischer Thanks: I thought I'd covered all the bases :(

@MarcelKrüger Do you have any concrete example where the superscript sits too high?

TL list is ... interesting

@mickep Take XITS in cramped style: It looks basically the same as the normal style instead of being cramped.

\setupbodyfont[xits]
\starttext
$$
  \displaystyle x^2
  \crampeddisplaystyle x^2
$$
\stoptext

@MarcelKrüger Thanks, I will look at the example tomorrow. It is much easier with an example.

@JosephWright A bit saddening, isn't it? (I guess you refer to a recent thread.)

@mickep Yes and yes

@mickep Sure, but my question is less about the actual behavior and more that the use of parameters doesn't seem to make sense. Increasing one parameter makes some superscripts higher and others lower which just seems weird.

@JosephWright -- Don't you mean "off the rails"?

@barbarabeeton Another way of puitting it