chiark / gitweb /
 author Ian Jackson Sun, 13 May 2012 12:59:43 +0000 (13:59 +0100) committer Ian Jackson Sun, 13 May 2012 12:59:43 +0000 (13:59 +0100)
 article.tex patch | blob | history strategy.tex patch | blob | history

index 18c75df..eb490e6 100644 (file)
\newcommand{\pcy}{\pay{C}}
\newcommand{\pcn}{\pan{C}}

+\newcommand{\pd}{\pa{D}}
+\newcommand{\pdy}{\pay{D}}
+\newcommand{\pdn}{\pan{D}}
+
\newcommand{\pl}{\pa{L}}
\newcommand{\ply}{\pay{L}}
\newcommand{\pln}{\pan{L}}
index 6fb4b4f..d9a2545 100644 (file)
@@ -91,7 +91,7 @@ We run the following algorithm:
\item Set $\allpatches = \{ \}$.
\item Repeatedly:
\begin{enumerate}
-\item Clear out the graph $\hasdirdep$ so it has neither nodes nor edges.
+\item Clear out the graph $\hasdirdep$ so it has no edges.
\item Execute {\bf Rank-Recurse}($\pc_0$)
\item Until $\allpatches$ remains unchanged.
\end{enumerate}
@@ -99,21 +99,85 @@ We run the following algorithm:

{\bf Rank-Recurse}($\pc$) is:
\begin{enumerate}
+
+\item If we have already done {\bf Rank-Recurse}($\pc$) in this
+ranking iteration, do nothing.  Otherwise:
+
\item Add $\pc$ to $\allpatches$ if it is not there already.
-\item Let $\set S_{\pcn} = h(\pcn) + +\item Let +$$+ \set S = h(\pcn) \cup \bigcup_{\p \in \allpatches} \bigcup_{H \in h(\pn) \lor H \in h(\py)} - \{ \baseof{E} \; | \; E \in \pendsof{H}{\pcy} \} + \{ \baseof{E} \; | \; E \in \pendsof{H}{\pcy} \} +$$ and$W = w(h(\pcn))$-We write$\set S = \set S_{\pcn}$where unambiguous. -\item While$\exists_{S \in \set S} S \ge W$: +\item While$\exists_{S \in \set S} S \ge W$, +update$W \assign S$and$\set S \assign \set S \, \backslash \{ S \}$+ +(This will often remove$W$from$\set S$. Afterwards,$\set S$+is a collection of heads to be merged into$W$.) + +\item Choose an order of$\set S$,$S_i$for$i=1 \ldots n$. + +\item For each$S_i$in turn, choose a corresponding$M_i$+such that $$+ M_i \le S_i \land \left[ + M_i \le W \lor \bigexists_{S_i, j<i} M_i \le s_i + \right] +$$ + +\item Set$\Gamma = \depsreqof{W}$. + +If there are multiple candidates we prefer$M_i \in \pcn$+if available. + +\item For each$i \ldots 1..n$, update our putative direct +dependencies: +$$+\Gamma \assign \text{\bf set-merge}\left(\Gamma, + \left[ \begin{cases} + M_i \in \pcn : & \depsreqof{M_i} \\ + M_i \not\in \pcn : & \{ \} + \end{cases} \right], + \depsreqof{S_i} + \right) +$$ + +\item Finalise our putative direct dependencies +$
+\Gamma \assign g(\pc, \Gamma)
+$+ +\item For each direct dependency$\pd \in \Gamma$, + +\begin{enumerate} +\item Add an edge$\pc \hasdirdep \pd$to the digraph (adding nodes +as necessary). +If this results in a cycle, abort entirely (as the function$g$is +inappropriate; a different$g$could work.) +\end{enumerate} +\item Run${\text{\bf Rank-Recurse}}(\pd)$. -Update$W \assign S$and$\set S \assign \set S \, \backslash \{ S \}$\end{enumerate} +The results of the ranking phase are: + +$ \allpatches, \hasdirdep $and hence the completion of$\hasdirdep$+into the partial order$\hasdep$. + +For each$\pc$, the base branch starting point commit$W_{\pcn} = W$, +the direct dependencies$\Gamma_{\pc}$, +the ordered set of base branch sources$\set S_{\pcn} = \set S,
+S_{\pcn,i} = S_i$+and corresponding merge bases$M_{\pcn,i} = M_i\$.
+
+
+
\section{Planning phase}

The results of the planning phase consist of: