asymmetric spacing around \sideset

[davidcarlisle]

Brief outline of the bug

The spacing around \sideset in amsmath is asymmetric and could be fixed by forcing \mathop interpretation at the left hand side see @eg9 's analysis at

https://tex.stackexchange.com/q/412575

Minimal example showing the bug

\RequirePackage{latexbug}
\documentclass{article}
\usepackage{amsmath}


\begin{document}



\[
\left|\sideset{^x}{}\sum\right|
+
\left|\sideset{}{}\sum_{x<n\le x+y}\right|
+
\left|\sideset{}{^x}\sum_{x+y<n\le x+y}\right|
\]
\makeatletter
\renewcommand{\sideset}[3]{%
  \@mathmeasure\z@\displaystyle{#3}%
  \global\setbox\@ne\vbox to\ht\z@{}\dp\@ne\dp\z@
  \setbox\tw@\box\@ne
  \@mathmeasure4\displaystyle{\copy\tw@#1}%
  \@mathmeasure6\displaystyle{#3\nolimits#2}%
  \dimen@-\wd6 \advance\dimen@\wd4 \advance\dimen@\wd\z@
  \mathop{\hbox to\dimen@{}}\!%
  \mathop{\kern-\dimen@\box4\box6}%
}
\makeatother
\[
\left|\sideset{^x}{}\sum\right|
+
\left|\sideset{}{}\sum_{x<n\le x+y}\right|
+
\left|\sideset{}{^x}\sum_{x+y<n\le x+y}\right|
\]

\end{document}

(see log analysis at the above link)

[FrankMittelbach]

looks like a clear bug to me that is worth fixing. Would it be advisable (to start) using latexrelease within the amsmath code to be able to do such fixes rather than introducing strange "options" to fix stuff?

[davidcarlisle]

I am not sure latexrelease really does the right thing here, if you have an existing document that has micro positioning fixes assuming the old spacing then you may want to keep that without forcing your entire latex processing for all packages and the kernel is reverted to 2017 I think using latexrelease in a package only really works if you need the package code to target different versions of the kernel.

…

On 29 January 2018 at 09:43, Frank Mittelbach ***@***.***> wrote: looks like a clear bug to me that is worth fixing. Would it be advisable (to start) using latexrelease within the amsmath code to be able to do such fixes rather than introducing strange "options" to fix stuff? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#14 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABNcAmddw9PMwnWXdW3MuHeyTTBDS_HAks5tPZLZgaJpZM4RwRHs> .

[FrankMittelbach]

not sure why. If I have written a paper in 2017 assuming \sideset in its old definition and I reprocess it and it comes out wrongly I could add latexrelease from that date when I wrote the paper and that should result in using code as if it was 2017. Of course if there are other packages not using latexrelease and changing the formatting then it could still come out differently (but then due to those other package and it would come out differently regardless of \sideset fixed or not). Implementing fixes by providing options to revert has the problem that it is either adhoc or your essentially implement a "amsmathrelease" method. Why that should be better than simply using latexrelease is not clear to me.

[davidcarlisle]

It depends... If for example you are using luatex and the latex kernel has changed to make it work in an updated luatex engine, then reverting all packages to an old state may not work at all. (In theory the kernel itself should be Ok as we don't mark such "forced" changes with latexrelease) but reverting the entire format (and any third party packages that are testing latexrelease) just to get the old behaviour for one construct seems a rather dangerous sledgehammer to be wielding.

Agreed it avoids a proliferation of ad hoc commands or options to revert specific changes but....

[wspr]

Considering how long it has been since amsmath has seen “breaking” changes, how about a single [compat] option that rolls back to the semi-frozen status of the 2000s?

[eg9]

Unfortunately, the proposed fix doesn't really work in general. The logic under \sideset is weak to begin with.

\documentclass{article}
\usepackage{amsmath}
\begin{document}
\[
\left|\sideset{}{^\text{quite very very very long}}\sum_{aaaaaaaaaaaaaaaaaa}\right|
\]
\end{document}

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.

[stale]

This issue has been automatically marked as stale because it has not had recent activity.