@MarcelKrüger I see what you mean. But it seems to me that it is xits that is just odd. For all other fonts (that I tried) the adjustment in cramped mode seems to work. While testing a bit, it seems that Look for example what ` \dm{\left(\vrule height 3mm depth 0mm \vrule height 0mm depth 5mm\right)^2_3}` gives in xits and stixtwo:
Either we miss something about xits, or the font is very broken.
@MarcelKrüger Regarding the power 2 of x that you show: I have now tried a bit more to set xits up, and it is the font parameter SuperscriptBaselineDropMax that is too large. It is 375 in the font. If I set it to 175, the cramped power looks ok. There are several things in play, here.
Left uncramped, right cramped. Top xits, bottom latin modern.
I'd wish the TL list would punish everybody who starts yet again a "zoom or not to zoom" conversation with having to read all future messages in comic sans!
@mickep Of course this is caused by SuperscriptBaselineDropMax, but my point was that it doesn't make sense that SuperscriptBaselineDropMax has this effect. The parameter is defined in OpenType as: "Maximum allowed drop of the baseline of superscripts relative to the (ink) top of the base. [...] Positive for superscript baseline below the base top." So a higher value here should make the superscript lower, not higher.
Of course if it is set to a value lower than SuperscriptShiftUp the changed behavior doesn't have any effect, but I don't see why it would be weird for a font to want such relatively low exponents. And especially not why the engine should put them especially high in response.
@mickep Sorry, I wrote that when I was a bit frustrated for independent reasons... The "of course" was not supposed to mean that it's generally obvious what is going on but only that the whole thing started with me writing Hand that I think that it seems odd that SuperscriptBaselineDropMax sometimes overwrites SuperscriptShiftUp for single characters, therefore it didn't come as a big surprise to me that changing SuperscriptBaselineDropMax would change this.
@MarcelKrüger Oh, no worries. That was merely a comment that to me the specification is not really clear. I will have to read it through to understand more in detail, in particular if there are more rules than one that need to be tried for and to understand the logical order.
@Werner I believe @barbarabeeton is working through them (or we could dig out the old scripts for \\ debacle which got edits down to 5s per post and could do a few thousand at a time)
@Werner -- As @DavidCarlisle says, I've been chipping away at the backlog. I've still got a list of about 30-40. And it's sometimes not just the url tat needs to be changed, I'll go and do another batch now, since I've been reminded. What's the question number of the one you've gotten hung up on? (I can't change comments though.)
@barbarabeeton Oh, okay... I thought there would be more based on the moderation flag description. Here's the one in question: tex.stackexchange.com/a/9868/5764
30-40 is really not a lot given there's ~250k questions and ~300k answers. So it seems reasonable to just tackle the remaining bunch manually (in my opinion).
@Werner -- Okay, fixed that and a bunch more. Actually, the remainder of my list was only 20 items, so it's now down to only one more pass. Then I will ask @DavidCarlisle to help by showing me how to run a check at the input text level, in case I missed anything. (I'm not very good at that; he has shown me before, but I've lost the instructions.)
@DavidCarlisle -- Thanks. I don't know when I made my list, but it stopped at 418001, so anything later wouldn't be on it. Also, I think I checked only at "surface" level, looking for "UK TeX", so i clearly wouldn't have gotten everything. Yes, looking for the actual url is best. Since I've actually changed text in nearly all the posts I've edited, an automatic update wouldn't do that, so I'll just keep on slogging. Sigh. But why is anyone still using tex.ac.uk in 2022???
@barbarabeeton not sure when the dns finally got turned off, looking at my mail looks like november 2021 we decided not to renew I guess it didn't last much longer
@DavidCarlisle -- Thanks for the pointer. Unfortunately, most occurrences also are best with some (small) changes in the text, and those are all different. I'll think about this; There may be some reasonable compromise.
@DavidCarlisle -- That's what I remember too. There's a big gap, from 2018-06 to 2022-07. Oh. I just looked at the last two (650855 and 656477), and the url is in a code example comment. Oh, well, I've been fixing those too. But ... those were both posted by @samcarter! Oops!
@DavidCarlisle -- Your second list really is probably better (if bigger). The second hit points to the tex,ac,uk CTAN mirror, and I never would have thought of looking for that ... but it does need to be changed to something that can be found.
@barbarabeeton er no:-) Anyone here actually know any sql? if select Id as [Post Link],Id gives two colums post link and Id, why is select Id as Id,[Post Link] a syntax error?:-)