Hi!
I'm imitating the argument-parsing of git-rebase(1) in a shell script. For the <branch> argument, if the user specifies a branch name, we should use the name; if it specifies something else, we should use that *as a detached HEAD*; and if it doesn't specify anything, we should use the active branch (if there's one), or the detached HEAD (else).
I've implemented it this way:
if test $# -lt 1; then git rev-parse --abbrev-ref HEAD; git rev-parse HEAD; else git rev-parse --verify --symbolic "$1"; git rev-parse --verify "$1"; fi \ | sed '/^HEAD$/d' \ | sed '1!d' \ >"$dir/head-name";
I wonder if I'm overthinking and maybe there's a simpler way. Or maybe this is the way.
The script above works by defaulting to HEAD if not specified, or using the commit ID in $1 (I've shifted, so it's just the last parameter) if specified. In the case where HEAD is explicitly specified, the user wants a detached HEAD, while if we've done it, we want to try the active branch if any.
The sed(1) part (together with the second commmand in each branch) makes sure we never keep the literal HEAD, but rather a reference or a hash.
Is this the simplest way to do it, or is there a better way to do it?
Have a lovely day! Alex
-- <https://www.alejandro-colomar.es>