12:33 AM

4 hours later…
4:13 AM
@PauloCereda Stupid.

1 hour later…
5:22 AM
@Canageek Example?

6:15 AM
An update of fancyhdr!! :-)

6:42 AM
@DavidCarlisle ou missed one instance of TaXworks :-)

@Johannes_B was just fixing that:-) (the mis-spelling being the only understandable part of the question)

@DavidCarlisle :-)

7:15 AM
@PauloCereda vvvv
For those who wondered what the pic I posted last night was saying, I translated it..though it may confuse you more. https://t.co/cZUng9TTsK

7:30 AM
@Johannes_B oh my
@AlanMunn oh my
@AlanMunn I woudn't say I didn't see it coming, as Apple came up with the USB-C thingy in the newer Macbooks.
Also, Apple wants to kill its iPod line, as they are (already too much) overpriced with little to none features added. Only Nano, Shuffle and Touch are available, with no updates in the last 4 years. It seems they want people to move from iPods to iPhones.

7:58 AM
@JosephWright @Werner Just noticed that the voting correction indirectly affected others: tex.stackexchange.com/users/41885/subham-soni?tab=reputation

@Johannes_B why "indirect" that looks fairly direct to me?

@DavidCarlisle The green tikmark is still there. If it is indirect or direct ... no idea.

8:13 AM
@Johannes_B The second in a few days

8:28 AM
@TorbjørnT. He uses TeXstudio. He has already asked the same question... well, slightly different.

@Alenanno Ah, I see. That's the reason for the "question modified" in the title then.

@TorbjørnT. He deleted it, though.
@TorbjørnT. Yes, eheh :D

and the undeserved \sum tick goes to ....

@DavidCarlisle What's that?

1

I would like the text, that is i=1 and N, on the summation to be top, and bottom. something like this I'm using the packages, amsmath, relsize, and physics. \label{eq: TEST} TEST = \mathlarger{\frac{{\mathlarger{\sum}}_{i=1}^{N}\frac{\abs{M_{\mathrm{i}} - m_{\mathrm{i}}...

8:37 AM
@TorbjørnT. :D ahah I see

8:53 AM
Gotta love the Latex one. :D

9:04 AM
@percusse any idea how to solve the remaining problem, i mentioned in the comments? tex.stackexchange.com/questions/328411/… I guess, it should be possible to "map" a distance of 1 to 3 onto all of the plot points (like xtick, xticklabels, domain)?

9:44 AM
Written exam grading done: only 2 out of 7 qualified for the oral examination :(

@yo' A colleague showed me how 10 out of 11 students in second-year analysis couldn't manage to solve $\frac{y^4+\sqrt{y^2}}{\sqrt{y^2}}=0$ (it was “find the intersections with the axes of some domain”, so that equation resulted from setting $x=0$).

@egreg nothing as sophisticated, or well, maybe. It was tough for them to find the tangent line of the solution of $y'=\cos^2 y \cos x$ passing through the point $(x,y)=(0, \pi/4)$. Actually, none of the 7 I graded finished this.

@yo' A couple of them wrote $\sqrt{y^2}=|y|$ and then wrote, at the numerator, $y^4\pm y$

@egreg that's one of the good ways to go, isn't it? The other one is substitution $z=y^2$, $z\geq0$, but that's a troublemaker actually :)

@yo' Yes!!!! :D
@yo' Could they arrive at $y=\arctan(\sin x+c)$?

9:55 AM
@egreg one could. But this way leads to a disaster :D I mean, do you want to take the derivative of this, only to find out that the slope is $y'(0)=1/2$?

@yo' Of course

@egreg you're a masochist :D

1 hour later…
11:02 AM

2 days ago, by yo'

@DavidCarlisle Like being on the LaTeX team, really ...

@SeanAllred for news about xkcd or cricket, you need to be quicker than that...

11:44 AM
@yo' (if you "qualify" for oral exam here, it basically means you're doing it the 3rd time and will get kicked out if you fail. :))
my favourite, today.

11:57 AM
@DavidCarlisle I thought as much :(

@pogo-otter here both writtwn and oral are obligatory

@yo' is spelling important in the writtwn exam?

12:11 PM
Quack from SP!

@DavidCarlisle it's math. Of course spelling is important!
@PauloCereda SP? ... Spain? :)

@pogo-otter no:-)

@pogo-otter São Paulo. :)
The state capital.

@PauloCereda damn, so close.
@PauloCereda at least it's the same language.
(kidding.)

12:30 PM
0

I'm using Lyx and I want to include unnumbered chapters to the table of contents such as the introduction and conclusion. So, I used \addcontentsline{toc}{chapter}{Introduction} right after declaring the chapters. The problem is that I get this error: use of \reserved@a doesn't match its d...

@DavidCarlisle ^^^ Oh to make everything that needs it \protected

@JosephWright if you do, you'll find lyx uses tex2 underneath

@DavidCarlisle :-)

@JosephWright I bet you would find some publisher still using Y&Y TeX (which isn't etex)

@DavidCarlisle I'm sure
@DavidCarlisle As I've said, my suspicion is such people don't update their format

@JosephWright certainly

12:35 PM
@DavidCarlisle I think I also pointed out that the team did say e-TeX was required some time ago

@JosephWright yes but saying that distributions should build latex on etex not tex (rather than having separate latex and elatex formats as previously) isn't quite the same as making the code not load on classic tex.

@DavidCarlisle I guess so

@JosephWright I don't care if new features don't work, but I think the code should load, so if a previously fragile command is made robust via \protected\def\foo... I'd rather we had \let\protected\relax somewhere before that if etex isn't enabled)

@DavidCarlisle Fine, fine, I do understand (honest)

@JosephWright :-)
of course someone is bound to be testing for etex by \ifx\protected\@undefined.... but you can't win them all.

12:40 PM
@DavidCarlisle We probably should revisit this on the team list and the 2e news: there is a statement about e-TeX in issue 16 (2003/12/01, before I was using LaTeX) but nothing since
@DavidCarlisle I always do \begingroup\csname protected\endcsname\gdef ... or equivalent

@JosephWright oops after I'd stopped using latex regularly:-)

@DavidCarlisle \let\PROTECTED\protected\ifx\protected\undefined\let\protected\relax\fi ...\let\protected\PROTECTED for example

@JosephWright something like that, meanwhile I note that ltnews16 says etex is not required which isn't quite the same as saying it is required :-)

@DavidCarlisle Yes, I thought there was some more definite statement: I'll raise on the team list

1 hour later…
1:44 PM
@Johannes_B @JosephWright: Are you able to see the legit-ness of this?

@Werner I know why this has happened, yes

I'd imagine the only reason why this would be the case is if there is some sock-puppetry between the two users.
@JosephWright Oh, okay...

@Werner, @Johannes_B (others) There has been staff involvement here: it's not an 'accident' due to some algorithm. From what I and the other mods know, there are good reasons for the action.

@JosephWright Good to know.
@JosephWright: I would say my interest in this is spurred by the fact that it's a first.
For some, new events always cause interest...

@Werner We have had some other manual voting reversal, but perhaps you've not seen it

1:50 PM
This is the experience :)
Source: xkcd.com/242
@JosephWright That must have been tedious.

2:50 PM
@JosephWright seems like we are better at covering up Hans' changes than Hans is:-)
0

The newest versions of the package xy-pic does not work properly with LuaLaTeX, version 0.95.0. Consider the following simple example: \documentclass{article} \usepackage[T1]{fontenc} \usepackage{fontspec} \usepackage[all]{xy} \begin{document} \[ \xymatrix { A\ar[r]\ar[d] & B\ar[d]\\ C\ar[r]

@JosephWright actually no, I made the error of believing the OP: the context luatex-pdf.tex works as well as luatex85.sty here.

3:29 PM
@DavidCarlisle Surely not changing xypdf.tex

@egreg true but I didn't check until later if loading luatex-pdf.tex in the document fixed it (luatex85.sty is still better though:-)

4:05 PM
hey! have any of you ever typeset an obituary?

4:18 PM
@DavidCarlisle no unless you write the exam in a shaking tube

@naphaneal -- what's the question?

4:34 PM
@JosephWright it's london bus day:
! Use of \\setof doesn't match its definition.
\new@ifnextchar ...served@d = #1\def \reserved@a {

@barbara the question is, if you may have an template i could use, as I am currently in the situation of planning my kid brother's funeral. and I was thinking to make a card

@naphaneal -- sadly, no. i have only set obituaries as articles in tugboat, and that's no help. i'm very sorry for your loss.

@naphaneal oh my, I am so sorry. My condolences.

@barbara @Paulo thanks.

@DavidCarlisle :)

4:59 PM
Hy! I commented the following answer: http://tex.stackexchange.com/a/164186/22655
When I tried to compile the code from this answer the compilation failed. Details are in the comment. Anybody has an idea why?

@GillesBonnet A version issue? Add \listfiles to your input and give us the source file and resulting File list from the .log
@GillesBonnet For exactly the source in the answer I get
 *File List*
scrbook.cls    2016/06/14 v3.21 KOMA-Script document class (book)
scrkbase.sty    2016/06/14 v3.21 KOMA-Script package (KOMA-Script-dependent bas
ics and keyval usage)
scrbase.sty    2016/06/14 v3.21 KOMA-Script package (KOMA-Script-independent b
asics and keyval usage)
keyval.sty    2014/10/28 v1.15 key=value parser (DPC)
tocbasic.sty    2016/06/14 v3.21 KOMA-Script package (handling toc-files)
scrsize12pt.clo    2016/06/14 v3.21 KOMA-Script font size class option (12pt)

After adding \listfiles at the really begining of the source file I get this File list in the .log:

*File List*
scrbook.cls 2015/10/03 v3.19a KOMA-Script document class (book)
scrkbase.sty 2015/10/03 v3.19a KOMA-Script package (KOMA-Script-dependent ba
sics and keyval usage)
scrbase.sty 2015/10/03 v3.19a KOMA-Script package (KOMA-Script-independent
basics and keyval usage)
keyval.sty 2014/10/28 v1.15 key=value parser (DPC)
@JosephWright And she source is the one of the answer. I copied and pasted without changing anything

5:41 PM
@GillesBonnet Your system is obviously older, so if you want to try the code you should update / get a new texlive.

@UlrikeFischer and @JosephWright. Ok. I see. Thanks.

6:04 PM
@GillesBonnet Grab l3kernel.tds.zip, l3packages.tdz.zip and l3experimental.tds.zip from CTAN if you can't do a full install

6:40 PM
Is there a standard macro \aswideas{x}{y} which sets x pretending it is as wide as y, taking into account math modes, not adding extra space, and what not?

@MarianoSuárez-Álvarez in math mode? not so simple

yup, mathmode
I know :-)

7:04 PM
Why can't a TeXnician kidnap anybody? Because he'd cut the newspaper letters preserving ligatures.

7:15 PM
@MarianoSuárez-Álvarez yes (most likely) it depends exactly what you mean:-)
@MarianoSuárez-Álvarez do you mean this has to work within a math expression so x+ \cos \aswideas{x}{(x)} has to set x but with the spacing from the mathopen ( at the start of (x) ?

I'm trying to get a system of equations to be aligned nicely, for once :-)
I have a_{1,1} and a_{1,m} that should be aligned, and since the m is wider than the 1, what messes the whole thing up. I could use an array, but the easiest way would be to set either the second one in a_{1,1} as wide as an m, or maybe the whole a_{1,1} as wide as the a_{1,m}
@DavidCarlisle that's be too much, I guess. my x and y will generally not have special spacing magic line mathopens and such

@MarianoSuárez-Álvarez well if they are simply a mathord like 1 or m it's not a problem at all is it?

but in measuring the m one needs to typeset it in the same scriptscriptscript size as the one needed for the context the whole thing will end up. I don´t think I know how to do that

@MarianoSuárez-Álvarez although I'd be tempted to just use \def\foo#1{\makebox[3pt]{$\m@th\scriptstyle #1$}} then _{1,\foo{1} and _{1,\foo{m} that way you can make the used width a fraction less than the m if it looks better rather tahn having a very spaced 1

that gets you an extra mathsurround around the 1, no?

7:23 PM
:32192301

that's a good point

@MarianoSuárez-Álvarez not now:-)

heh
The Powers That Be should come up with a new macro for mathtools which does this :-)
With an optional [rlc] arg and all

@MarianoSuárez-Álvarez mathtools, that's @daleif , I just have amsmath:-)
@MarianoSuárez-Álvarez oh isn't it \mathmakebox (in mathtools)

you can then remove the tiny extra space before multlined, then! :-)

7:27 PM
@MarianoSuárez-Álvarez NO, unfortunately.

@DavidCarlisle the problem is in measuring, though

@MarianoSuárez-Álvarez which automatically adds \scriptstyle or not, depending

oh it does?

@yo' probably yes, actually.

@DavidCarlisle really?!

7:29 PM
@yo' was looking at it earlier in the year, it's probably safe to change it to say, if lastskip negative do nothing else do \!

@DavidCarlisle yeah, and at the end?!

@yo' the end is OK, that's the whole point it's not symmetrical, the bad space is on the left (as far as I remember)

@DavidCarlisle the tugboat paper does not explain exactly what that macro does, and the mathtools refers to it. I guess I should have looked at the code!

@DavidCarlisle ah, so it's different from aligned then?

@MarianoSuárez-Álvarez ohh documentation didn't occur to me to look for that, I just looked down the stripped mathtools.sty in emacs:-)
@yo' probably not, I'll check later;-)

7:41 PM

@DavidCarlisle the code for vdotswithin might be handy, but for systems I'd just use a load of &'s, perhaps even make some macro to build it for me

@daleif in my case, it was not really a system of equations but one big expression with some structure, not really a candidate for automatization

8:02 PM
@yo' vvvvv
\documentclass{article}
\usepackage{amsmath}
\usepackage[T1]{fontenc}

\begin{document}

|\begin{aligned}a\end{aligned}|

|$a$|

|\!\begin{aligned}a\end{aligned}|

|\!\begin{aligned}a\end{aligned}\!|

\end{document}
@yo' the correction just needs to be at the start, and (reasonably safely) just if there isn't already negative space there

@DavidCarlisle yeah, sounds reasonable to me

@yo' ah now I remember the unreasonable part.

@MarianoSuárez-Álvarez -- rather than actually measuring, why not just take advantage of an \hphantom (using the wider element) and a \mathrlap to plop the narrower bit into that space.

@yo' but don't let @barbarabeeton know we are discussing amsmath, it makes her nervous.

haha
@barbarabeeton Ah.

8:09 PM
@DavidCarlisle -- oh, i've got worse things to make me nervous at the moment, believe it or not. (but amsmath is there as a close second.)

@yo' if you put \showthe\lastskip at the start of \aligned then it will show 0pt or -3mu depending if \! is there, but if you try to use \lastskip as in \muskip0=\lastskip it tells you that they are incompatible units so i don't know a good error free way to detect the space before a math construct

@DavidCarlisle a "bug" ("feature") in Knuth's TeX? :-)

@DavidCarlisle didn't egreg also have an example where one needed \!  inside aligned? (special case, though I needed it at one point)

I can't test just now, but what would \ifdim\lastskip<-2.5mu do?

@yo' which reminds me I didn't get a cheque for the overhanging halign spans discussed here the other day that was a documented feature as well (if you squinted hard enough at the texbook)
@yo' it dies

8:14 PM
@DavidCarlisle goodness. And \tempskip=\lastskip also dies?

@yo' it dies

@DavidCarlisle goodness! :-)
So, something like \setbox0=\hbox{$\mkern-\lastskip$} \the\wd0 ?

@yo' basically there are not any constructs that will take a skip or a muskip, tex lets you off in \showthe but \lastskip really wants to be a skip
! Missing number, treated as zero.
$l.10 \setbox0=\hbox{$\muskip\lastskip @daleif -- actually, the presence of \, at the beginning of align... structures is a well known "problem". (well, well known to anyone who has had to dig through the documentation as many times as i have; no, the reason given is not unambiguous.) incompatibility between \muskip and \lastskip i didn't know about. @DavidC -- maybe a question for dek (but no answer until 2020). @yo' ^^^ 8:18 PM @DavidCarlisle \expandafter\mkern\the\lastkip or alike? @yo' nope, I would guess no amount of expandafters will work, see if there isn't a skip there, \lastskip is 0pt and \mkern0pt is an incompatible units error.... @DavidCarlisle ah damn, it's 0pt and not 0mu well, then, redefine \, ;-) @yo' ask @JosephWright those pesky units are important. @yo' no:-) @DavidCarlisle twas a joke @yo' I know @yo' afterpage was a joke but hasn't stopped people using it for 20 years 8:21 PM @DavidCarlisle :) @yo' actually you are on to something there..... @DavidCarlisle \expandafter\pdfstrcmp\expandafter{\the\lastskip}{-3mu} ? @yo' yes basically \the like \showthe works, so you pick up the string see if it is mu or pt then do the right thing (remember \lastskip removes, so you need to put back whatever it was using \hskip or \mskip, depending) @DavidCarlisle yeah, the "removes" part is slightly problematic. Could there be a glue? (Are there glues in mathmode?) @yo' yes hskip and muskip are both glue @yo' \! is -\thinmuskip which happens not to stretch with its default value but: \thinmuskip=3mu \medmuskip=4mu plus 2mu minus 4mu \thickmuskip=5mu plus 5mu 8:33 PM @DavidCarlisle then you could do something like \afterassignment\foo\tempcntr\the\lastskip\endcsname where \def\foo{\csname choosesuffix} and \def\choosesuffixpt{WHATEVER WITH PT} and \def\choosesuffixmu{WHATEVER WITH MU} but I think this wouldn't work with plus minus @yo' no need to use a counter you can use a skip or muskip register having looked in advance to see if there is an mu or pt at the end of the expansion of \the @DavidCarlisle yeah I know :-) Well, you're the master of these things... @yo' let's see what I can break. @PauloCereda's thesis probably has some aligned in it.... does one really need to turn the \lastskip into a string to do that? :-| @MarianoSuárez-Álvarez yes (I think so) 8:37 PM @DavidCarlisle only if it's got characters already :-) @MarianoSuárez-Álvarez since there's an allowed type mismatch there, yes. @yo' for some reason I was giving him the benefit of the doubt this evening @yo' oh of course, \thelastskip doesn't remove lastskip is only removed if you make an assignment, \skip0=\lastskip, that makes it easier... ! You can't use \relax' after \the. <inserted text> \relax l.13 1|\mskip5mu \zz
WHO WAS IT THAT PUT THE \relax THERE ???

@DavidCarlisle WHERE "THERE"?!

@yo' doesn't really matter (in an \if- test actually) , just cross the stupid thing added \relax first (no previous error) then generated an error that \relax isn't allowed:-)

@DavidCarlisle :-) And I thought you've got the full TeX interpreter implanted into your brain :-)

8:52 PM
@yo' this adds \! (as [\!] so you can see it) just if it is not preceded by a negative skip or muskip, so should be safe to put at start of aligned without making it have too much negative space and over-print the preceding thing
\documentclass{article}
\usepackage{amsmath}
\usepackage[T1]{fontenc}

\makeatletter
\def\zz{%
\edef\foo{\expandafter\@car\the\lastskip\@nil}%
\if-\foo\else[\!]\fi}

\begin{document}

1|$\mskip5mu \zz$|

2|$\mskip0mu \zz$|

3|$\zz$|

4|$\mskip-5mu \zz$|

5|$\hskip5em \zz$|

6|$\hskip-5em \zz$|

\end{document}

@DavidCarlisle cool. I would just say that only \! shall be corrected, not arbitrary negative space

@yo' no that's backwards, you want to correct almost always (eg the usual case of \begin{aligned} you just need to not correct if there is a suspicion the correction already done \documentclass{article} \usepackage{amsmath} \usepackage[T1]{fontenc} \makeatletter \def\zz{% \edef\foo{\expandafter\@car\the\lastskip\@nil}% \if-\foo\else\!\fi} \makeatother \begin{document} 1|a$| 2|$\!\begin{aligned}a\end{aligned}$| 3|$\,\begin{aligned}a\end{aligned}$| 4|$\begin{aligned}a\end{aligned}| \let\zzz\aligned \def\aligned{\zz\zzz} 1|a$| 2|$\!\begin{aligned}a\end{aligned}$| 3|$\,\begin{aligned}a\end{aligned}$| 4|$\begin{aligned}a\end{aligned}| \end{document} @DavidCarlisle yeah, but think of tweaks like \begin{multline}\kern-3em\begin{aligned}... (been there done that I think) You'll shift me off by 1.5mu @barbarabeeton ^^^^ aligned without spurious space on the left ^^^^ :-) @yo' exactly. the above (should) detect that and so not add any correction (it needs to check \lastkern as well as \lastskip, will add....) @DavidCarlisle seen the minus sign in the kern? 9:04 PM @yo' ? @yo' yes \kern-3em so I wouldn't want to add any more negative space in that case. @DavidCarlisle I have used negative kerns in places like this before to fit things in where I needed them @DavidCarlisle ah right, stupid me... @yo' so any such use would not be affected by this change @DavidCarlisle I got lost in what's the old feature and what's the new feature reverting the old feature :-) @DavidCarlisle -- what i'd really like to see is the reason for the \, at the beginning of aligned documented in the .dtx file. (when you've updated amsmath, have you been updating the .dtx file at the same time?) maybe i can get agreement here that the \, can simply be removed; since we do keep records of what versions of packages are used for every job, and always rerun with the same versions, that shouldn't affect us. as for other users, ... @barbarabeeton you do, but many others do not 9:12 PM @yo' -- yes, i know that. (but many others also claim that they never have to rerun anything <some length of time> later. of course, i don't believe it ...) so here cases 1,2,5,6 unchanged by the addition, cases 3 and 4 get a new \! inserted: anyway, I gotta go, see you another time! \documentclass{article} \usepackage{amsmath} \usepackage[T1]{fontenc} \makeatletter \def\zz{% \edef\foo{\expandafter\@car\the\lastskip\@nil}% \if-\foo\else \edef\foob{\expandafter\@car\the\lastkern\@nil}% \if-\foob\else \!\fi\fi} \makeatother \begin{document} 1|a$| 2|$\!\begin{aligned}a\end{aligned}$| 3|$\,\begin{aligned}a\end{aligned}$| 4|$\begin{aligned}a\end{aligned}$| 5|$\mkern-5mu\begin{aligned}a\end{aligned}$| 6|\kern5pt$\kern-5pt\begin{aligned}a\end{aligned}| \let\zzz\aligned @DavidCarlisle -- i'm a little bit lost whether this is something that's intended to be put into amsmath to decide automatically whether the \, should be got rid of, or whether it's still a matter of eyeballing and applying the patch. @DavidCarlisle -- this looks much worse. @barbarabeeton no way we can just remove it I think, it would cause silent overprinting in existing documents that have a \! fix added. @barbarabeeton the former @barbarabeeton just look at case 4 before and after. The other cases are weird tests. 9:16 PM @DavidCarlisle -- that's what i was hoping. Is this an unreasonable comment? without any other evidence to the contrary, the most likely reason to get a \bfseries invalid in math mode error when you \cite something is that the citation text has \bfseries in math mode (which is an error) — David Carlisle 30 secs ago @DavidCarlisle -- case 4 is improved in the "after" version. are you sure that all the other examples are just weird? (my brain is just fried at the moment.) @barbarabeeton case by case: 1: no aligned, no change, 2: existing \! fix, no change, 3: positive space previously, small negative space added, 4: bare aligned, negative space added cancelling the internal \,, 5 and 6 already negative space before the aligned, so no change. @barbarabeeton so end result only 3 and 4 change, in 3 you get exactly the space you specfied before the alignment not thinmuskip extra and in 4 you get no space before the alignment instead of thinmuskip. @DavidCarlisle -- all make sense. thanks for explication. i'm trying to think of where i might find some examples here that could be used for a test to show the local interested parties. i'll let you know by mail. in the meantime, i think it seems like a good idea. 9:51 PM gad, what is this spate of "can't compile ..." after watching a "tutorial" on youtube! if i knew what to look for on youtube, i'd track down the perpetrator and kindly request that something be mentioned about "you do have to download and install tex live first". @DavidCarlisle vvvvv [paulo@cambridge documento] ack --tex "aligned"
conceitos.tex
503:\begin{aligned}
508:\end{aligned}

@PauloCereda no \! before them?

@DavidCarlisle Why should I? :)

@PauloCereda because your thesis examiner would be so concerned by a spurious thinmuskip to the left of the alignment you thesis would be failed as unreadable.
3

@DavidCarlisle Oh no!

10:16 PM
@HenriMenke Thanks. :)

10:33 PM
@PauloCereda For sure I'd notice it and reject the utterly wrong document.

@egreg Oh. :(

10:44 PM
@egreg more seriously what do you think of the above suggested change to aligned?

@DavidCarlisle Too many documents already rely on the presence of \,; removing it will do much damage.
@DavidCarlisle A new version of amsmath, with differently named environments…

@egreg that's a possibly valid viewpoint (and why it's not been fixed so far of course:-) I could of course make it an opt-in package option to not have the space, in which case could probably simply not have it rather than have the above complicated tests for lastskip tests
@egreg mathtools? perhaps
@egreg if you come across a document where adding \! before aligned makes things worse would be a good test (I do not doubt that there are cases) not including cases that have negative space already before aligned.

@DavidCarlisle An option could be acceptable, but several classes load amsmath by default, and…

@egreg oh I'm an expert at citing compatibility concerns as a reason for leaving bugs unfixed for decades, you don't need to help me to find excuses. I'm just wondering if I should be more modern and pro-active in my old age and fix things that are wrong...

@DavidCarlisle I think the space doesn't really get noticed in cases such as \left\(\begin{aligned}...\end{aligned}\right) and maybe it's even good there. Adding \, was an error to begin with.
We can blame Mike Spivak

10:56 PM
@egreg yes then Frank for not fixing it then Michael for not prodding Frank to fix it. So it's Ok it's not my fault.

@DavidCarlisle Once in a while it happens. :P

@egreg perhaps I should make the change, break a few documents then it would be my fault and order will have been restored. Seriously I don't know, having space on the left and not the right is a pretty clear straight bug, it's a shame if it's really not possible to do anything about it.

@DavidCarlisle is \usepackage{fixamsmath}` a solution?