20.6.1 How display-buffer works

The display-buffer command (as well as commands that call it internally) chooses a window to display by following the steps given below. See Choosing a Window for Display in The Emacs Lisp Reference Manual, for details about how to alter this sequence of steps.

• First, check if the buffer should be displayed in the selected window regardless of other considerations. You can tell Emacs to do this by adding the desired buffer’s name to the list same-window-buffer-names, or adding a matching regular expression to the list same-window-regexps. By default, these variables are nil, so this step is skipped.
• Otherwise, if the buffer is already displayed in an existing window, “reuse” that window. Normally, only windows on the selected frame are considered, but windows on other frames are also reusable if you change pop-up-frames (see below) to t.
• Otherwise, optionally create a new frame and display the buffer there. By default, this step is skipped. To enable it, change the variable pop-up-frames to a non-nil value. The special value graphic-only means to do this only on graphical displays.
• Otherwise, try to create a new window by splitting the selected window, and display the buffer in that new window.

  The split can be either vertical or horizontal, depending on the variables split-height-threshold and split-width-threshold. These variables should have integer values. If split-height-threshold is smaller than the selected window’s height, the split puts the new window below. Otherwise, if split-width-threshold is smaller than the window’s width, the split puts the new window on the right. If neither condition holds, Emacs tries to split so that the new window is below—but only if the window was not split before (to avoid excessive splitting).

• Otherwise, display the buffer in an existing window on the selected frame.
• If all the above methods fail for whatever reason, create a new frame and display the buffer there.