Sans-serif partial derivative symbol?

Question

How can I typeset a partial derivative symbol (∂) in an upright, sans-serif font?

I am using pdfLaTeX. I know this problem could be solved easily by switching to another engine — and feel free to post XeLaTeX or LuaLaTeX solutions for the benefit of posterity if you wish — but I am specifically looking for a solution within pdfLaTeX.

---

Edited to add an image to contrast what LaTeX generates with what I want:

Answer 1

Partial solution.

\usepackage{graphicx}
\renewcommand*\partial{\textsf{\reflectbox{6}}}

Answer 2

\documentclass{article}
\usepackage[T1]{fontenc}
\usepackage{kpfonts}
\def\uppartial{{\mathversion{sf}\ensuremath{\partialup}}}
\begin{document}
$\partial$ partial derivative symbol in \LaTeX

\sffamily
\uppartial{} partial derivative symbol in \LaTeX
\end{document}

You can also use \usepackage[partialup]{kpfonts} if you do not need the default version. Then \partial is upright by default.

Answer 3

How about using cmbright to get sans serif math and rotating the \partial?

\documentclass{article}
\usepackage{cmbright}
\usepackage{graphicx}
\newcommand{\upartial}{\rotatebox[origin=c]{15}{\ensuremath{\partial\mkern-2mu}}}
\begin{document}
\[\frac{\partial x}{\partial t}\quad\textnormal{vs.}\quad\frac{\upartial x}{\upartial t}\]
\end{document}

Edit:

One can get cmbright only in math mode with the following code. However, in doing so I discovered this solution is not very robust. As you can see in the image, the upright \partial gets typeset in \displaystyle despite being inside \(...\).

Code:

\documentclass{amsart}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc}
\usepackage{lmodern,cmbright}
\renewcommand*\familydefault{\rmdefault}
\usepackage{graphicx}
\newcommand{\upartial}{\mathsurround=0pt \rotatebox[origin=c]{15}{$\partial\mkern-2mu$}}
\begin{document}
Normal vs.\ upright partial: \(\frac{\partial x}{\partial t}\) vs.\ \(\frac{\upartial x}{\upartial t}\)
\end{document}

Note: I've also changed the definition of \upartial after reading @egreg's answer here, but the inline math problem persists.