For what it is worth. This adds a plot mark and respects mark options
.
\documentclass[margin=3cm]{standalone}
\usepackage{tikz}
\begin{document}
\begin{tikzpicture}[cmark/.style={append after command={plot[only marks,mark=#1] coordinates {(\tikzlastnode)}}}]
\coordinate[label={[text=red,left]:A},cmark={*,mark options={fill=white}}]
(A) at (0,2);
\path (2,2) node[circle,inner sep=1.5pt,draw,label={[text=red,left]:B}] (B){} ;
\draw[stealth-stealth] (A) to[out=90,in=90] (B);
\end{tikzpicture}
\end{document}

Note, however, that I do not see any advantage compared to option B
in which a node is used. Rather, I see a disadvantage since, if you connect either of them with an arrow, only the node knows the boundary, as illustrated. (Of course, in the node case you can connect to the center, too, by using B.center
instead of B
in the arc path.)
\coordinate[label={[text=red,left]:\pgfuseplotmark{*}A}] (A) at (0,2);
? Or\coordinate[label={[text=red,left]:A},label={[anchor=center,blue]:\pgfuseplotmark{*}}] (A) at (0,2);
? Please make your question clearer.\begin{tikzpicture}[cmark/.style={label={[anchor=center]:\pgfuseplotmark{#1}}}] \coordinate[cmark=*,label={[text=red,left]:A}] (A) at (0,2); \end{tikzpicture}
.dotlablel
macro with TikZ?”. ;-)node[circle,fill,inner sep=1.5pt]
). The latter has all the node anchors, i.e. if you connect it with an arrow, it looks good, but the plot mark stuff doesn't.