On Sun, Oct 28, 2018 at 01:50:25PM +0100, Anders Waldenborg wrote: > This new format placeholder allows displaying only a single > trailer. The formatting done is similar to what is done for > --decorate/%d using parentheses and comma separation. Displaying a single trailer makes sense as a goal. It was one of the things I considered when working on %(trailers), actually, but I ended up needing something a bit more flexible (hence the ability to dump the trailers in a parse-able format, where I feed them to another script). But your ticket example makes sense for just ordinary log displays. Junio's review already covered my biggest question, which is why not something like "%(trailers:key=ticket)". And likewise making things like comma-separation options. But my second question is whether we want to provide something more flexible than the always-parentheses that "%d" provides. That has been a problem in the past when people want to format the decoration in some other way. We have formatting magic for "if this thing is non-empty, then show this prefix" in the for-each-ref formatter, but I'm not sure that we do in the commit pretty-printer beyond "% ". I wonder if we could/should add a a placeholder for "if this thing is non-empty, put in a space and enclose it in parentheses". > diff --git a/Documentation/pretty-formats.txt b/Documentation/pretty-formats.txt > index 6109ef09aa..a46d0c0717 100644 > --- a/Documentation/pretty-formats.txt > +++ b/Documentation/pretty-formats.txt > @@ -211,6 +211,10 @@ endif::git-rev-list[] > If the `unfold` option is given, behave as if interpret-trailer's > `--unfold` option was given. E.g., `%(trailers:only,unfold)` to do > both. > +- %(trailer:<t>): display the specified trailer in parentheses (like > + %d does for refnames). If there are multiple entries of that trailer > + they are shown comma separated. If there are no matching trailers > + nothing is displayed. It might be worth specifying how this match is done. I'm thinking specifically of whether it's case-sensitive, but I wonder if there should be any allowance for other normalization (e.g., allowing a regex to match "coauthored-by" and "co-authored-by" or something). -Peff
