Documentation / technical / api-setup.txton commit Merge branch 'ao/p4-has-branch-prefix-fix' (3efeb51)
   1setup API
   2=========
   3
   4Talk about
   5
   6* setup_git_directory()
   7* setup_git_directory_gently()
   8* is_inside_git_dir()
   9* is_inside_work_tree()
  10* setup_work_tree()
  11
  12(Dscho)
  13
  14Pathspec
  15--------
  16
  17See glossary-context.txt for the syntax of pathspec. In memory, a
  18pathspec set is represented by "struct pathspec" and is prepared by
  19parse_pathspec(). This function takes several arguments:
  20
  21- magic_mask specifies what features that are NOT supported by the
  22  following code. If a user attempts to use such a feature,
  23  parse_pathspec() can reject it early.
  24
  25- flags specifies other things that the caller wants parse_pathspec to
  26  perform.
  27
  28- prefix and args come from cmd_* functions
  29
  30get_pathspec() is obsolete and should never be used in new code.
  31
  32parse_pathspec() helps catch unsupported features and reject them
  33politely. At a lower level, different pathspec-related functions may
  34not support the same set of features. Such pathspec-sensitive
  35functions are guarded with GUARD_PATHSPEC(), which will die in an
  36unfriendly way when an unsupported feature is requested.
  37
  38The command designers are supposed to make sure that GUARD_PATHSPEC()
  39never dies. They have to make sure all unsupported features are caught
  40by parse_pathspec(), not by GUARD_PATHSPEC. grepping GUARD_PATHSPEC()
  41should give the designers all pathspec-sensitive codepaths and what
  42features they support.
  43
  44A similar process is applied when a new pathspec magic is added. The
  45designer lifts the GUARD_PATHSPEC restriction in the functions that
  46support the new magic. At the same time (s)he has to make sure this
  47new feature will be caught at parse_pathspec() in commands that cannot
  48handle the new magic in some cases. grepping parse_pathspec() should
  49help.