Placement of cite:post in code flow

[moewew]

Consider the following MWE

\documentclass[british]{article}
\usepackage{babel}
\usepackage{csquotes}
\usepackage[backend=biber, style=apa, natbib=true, defernumbers=true]{biblatex}
\DeclareLanguageMapping{british}{british-apa}

\usepackage{filecontents}
\begin{filecontents}{\jobname.bib}
@book{OECD2009,
  title       = {P{ISA} {D}ata {A}nalysis {M}anual: {SPSS}, {S}econd {E}dition},
  publisher   = {OECD Publishing},
  year        = {2009},
  author      = {{Organisation for Economic Co-operation and Development}},
  address     = {Paris},
  shortauthor = {{OECD}},
}
\end{filecontents}
\addbibresource{\jobname.bib}
\addbibresource{biblatex-examples.bib}

\begin{document}        
  \cite{OECD2009,sigfridsson}

  \cite{OECD2009,sigfridsson}

\citereset\citeresetapa
  \cite{OECD2009}

  \cite{OECD2009}
\end{document}

You will find that in the first block the subsequent citation to the OECD source still triggers the full author (interestingly without the brackets - though that seems intentional with sabrackets?), while the second block correctly shows the shortauthor.

This seems to have to do with the placement of cite:post, if it is moved to the loopcode, every work triggers its use. With

\DeclareCiteCommand{\cite}
  {\usebibmacro{cite:init}%
   \usebibmacro{prenote}}
  {\usebibmacro{citeindex}%
   \usebibmacro{cite}%
   \usebibmacro{cite:post}}
  {}
  {\usebibmacro{postnote}}

the citations look as expected.

[moewew]

This came up again in https://tex.stackexchange.com/q/440475/35864

I'm quite confident that indeed only cite:post needs to be moved around a bit, but I'm not entirely sure, so I have not put in a pull request.

[moewew]

While I was investigating this I came across the definition of sabrackets again. It does an \ifciteseen

biblatex-apa/tex/latex/biblatex-apa/cbx/apa.cbx

Lines 186 to 205 in 0636af2

	\DeclareNameFormat{sabrackets}{%
	\ifciteseen
	{\usebibmacro{labelname:doname}%
	{\namepartfamily}%
	{\namepartfamilyi}%
	{\namepartgiven}%
	{\namepartgiveni}%
	{\namepartprefix}%
	{\namepartprefixi}%
	{\namepartsuffix}%
	{\namepartsuffixi}}
	{\mkbibbrackets{\usebibmacro{labelname:doname}%
	{\namepartfamily}%
	{\namepartfamilyi}%
	{\namepartgiven}%
	{\namepartgiveni}%
	{\namepartprefix}%
	{\namepartprefixi}%
	{\namepartsuffix}%
	{\namepartsuffixi}}}}

But the sbrackets format is only used if \cbx@apa@ifnamesaved is false and there is no sensible way that \ifciteseen is true, yet \cbx@apa@ifnamesaved is false (there actually is, because \cbx@apa@ifnamesaved has a reset option separate from the usual biblatex trackers), plus I think the output one would get in that case would likely be regarded as wrong anway. So I think sabrackets could be simplified to

\DeclareNameFormat{sabrackets}{%
  \mkbibbrackets{%
    \usebibmacro{labelname:doname}%
      {\namepartfamily}%
      {\namepartfamilyi}%
      {\namepartgiven}%
      {\namepartgiveni}%
      {\namepartprefix}%
      {\namepartprefixi}%
      {\namepartsuffix}%
      {\namepartsuffixi}}}

Coming to think of I think the entire sabracket thing could be simplified and I would probably go with

\documentclass[12pt,a4paper]{scrartcl}
\usepackage[T1]{fontenc}
\usepackage[utf8]{inputenc} 
\usepackage{lmodern}
\usepackage[ngerman]{babel}
\usepackage[backend=biber,style=apa]{biblatex}

\usepackage{filecontents}
\begin{filecontents}{\jobname.bib}
@online{BMG2018,
author = {{Bundesministerium für Gesundheit}},
shortauthor = {BMG},
title = {Gesund bleiben: Prävention und Gesundheitsförderung.},
date = {2018},
}

@booklet{BLS2011,
author = {{Bayerischer Landes-Sportverband e. V.}},
shortauthor = {BLS},
title = {Gesundheitssport im Verein},
date = {2011},
}
\end{filecontents}

\addbibresource{\jobname.bib}
\DeclareCiteCommand{\parencite}[\mkbibparens]
  {\usebibmacro{cite:init}%
   \usebibmacro{prenote}}
  {\usebibmacro{citeindex}%
   \usebibmacro{cite}%
   \usebibmacro{cite:post}}
  {}
  {\usebibmacro{postnote}}

\makeatletter
\newbibmacro*{cite}{%
  \iffieldequals{fullhash}{\cbx@lasthash}
% Multiple cites in one command
   {\setunit{\compcitedelim}%
    \usebibmacro{cite:plabelyear+extradate}}%
% Single cite
   {\ifthenelse{\ifnameundef{labelname}\OR\iffieldequalstr{entrytype}{patent}}
% No author/editor
     {\usebibmacro{cite:noname}%
       \setunit{\printdelim{nameyeardelim}}%
       \usebibmacro{cite:plabelyear+extradate}%
       \savefield{fullhash}{\cbx@lasthash}}
% Normal cite
     {\ifnameundef{shortauthor}
        {\printnames[labelname][-\value{listtotal}]{labelname}}%
        {\cbx@apa@ifnamesaved
          {\printnames{shortauthor}}
          {\printnames[labelname][-\value{listtotal}]{author}\addspace\printtext[brackets]{\printnames[sabrackets]{shortauthor}}}}%
      \setunit{\printdelim{nameyeardelim}}%
      \usebibmacro{cite:plabelyear+extradate}%
      \savefield{fullhash}{\cbx@lasthash}}}%
   \setunit{\multicitedelim}}
\makeatother

\DeclareNameFormat{sabrackets}{%
  \usebibmacro{labelname:doname}%
    {\namepartfamily}%
    {\namepartfamilyi}%
    {\namepartgiven}%
    {\namepartgiveni}%
    {\namepartprefix}%
    {\namepartprefixi}
    {\namepartsuffix}%
    {\namepartsuffixi}}

\begin{document}
\parencite{BMG2018, BLS2011}

\parencite{BLS2011}

\parencite{BMG2018}

\parencite{BLS2011} 

\printbibliography
\end{document}

because that gives slightly better results for #63

[plk]

Ok, can you make a pull request for this and I'll run the test docs.