1) It is fairly safe to assume that almost all edits to Emacs lisp
files will be done using the Emacs editor, so there is no need
to configure this in a way understood by other editors.
2) Trying to configure the intention of lisp code using editorconfig
causes the indention to be wrong. Here "wrong" does not mean, "it
uses tabs when everyone knows that spaces is the true path" (or
vice-versa), but in the sense of "in lisp indentation is subject to
the outer expression, and it has been like that for decades, but we
are just going to ignore that completely and pretend this is
actually python code".
For example, if we insert a new line character between the 1 and 2
in (progn (progn 1 2) then there are only two ways to intend that
correctly.
(We use "." to represent a leading space and "<-->" to represent
leading tabs. "<------>" for a tab when it is shown eight characters
wide or "<>" if only two.)
First correct way:
(setq-local indent-tabs-mode t)
(progn (progn 1
<------>......2))
Second correct way:
(setq-local indent-tabs-mode nil)
(progn (progn 1
..............2))
With the old editorconfig configuration,
[*.el]
indent_style = tab
indent_size = 2
max_line_length = 100
we get:
(progn (progn 1
<><><><>.2))
This is not how `progn' is indented. No special indentation rules are
defined for it so all arguments are supposed to be aligned. For
`prog1' however special indentation are defined and if we replaced the
second `progn' in this example with a `prog1', then this would
actually be correct. This just demonstrate that it is wrong to indent
everything the same for lisp; the reason that `progn' and `prog1' are
indented differently communicates different meanings to the reader.
I had faint hope that without setting `indent_size' things would work
correctly, but no,
[*.el]
indent_style = tab
max_line_length = 100
gives us:
(progn (progn 1
<------>.......2))
That doesn't even make sense if we pretend to be looking at python
code. Turns out that `indent_size' defaults to 2 even for lisp code
as can be demonstrated by additionally doing,
(setq-local tab-width 2)
which gives us:
(progn (progn 1
<>.......2))
which at least makes some sense.
Dirk-Jan C. Binnema
Jonas Bernoulli