[erlang-patches] [PATCH] Fix indentation of record fields in Emacs
Henrik Nord
henrik@REDACTED
Tue Jun 12 11:41:16 CEST 2012
Hi Tomas
Yes this is probably my bad that it is not included.
It has already been reviewed and approved and marked as answered so the
only thing missing is the Email notification for you and the list, and
the merger itself.
I will run it through the tests real fast.
Apologies for the wait, and thank you for the contribution!
/Henrik
On 06/11/2012 03:54 PM, Tomas Abrahamsson wrote:
> Hi,
>
> I never received any feedback on the patch to the emacs
> record indentation, it was not included in pu and not the R15B01.
> Was it good? Bad? Ugly? Or just forgotten?
>
> We've been using it extensively in a project with lots
> of records, and folks seem happy with it. We haven't
> seen any bad cases of indentation due to it.
>
> BRs
> Tomas
>
> On Sat, Feb 11, 2012 at 9:55 PM, Tomas Abrahamsson
> <tomas.abrahamsson@REDACTED> wrote:
>> In some situations, the indentation of record fields in Emacs was
>> strange. This example below shows how Emacs previously would indent
>> two similar pieces of code very differently:
>>
>> some_function_with_a_very_long_name() ->
>> #'a-long-record-name-like-it-sometimes-is-with-asn.1-records'{
>> field1=a,
>> field2=b}.
>>
>> x() ->
>> #some_record_name{
>> field1=a,
>> field2=b}.
>>
>> This changes the indentation to be like below for both cases:
>>
>> some_function() ->
>> #some_record{
>> field1=a,
>> field2=b}.
>> ---
>> lib/tools/emacs/erlang.el | 58 +++++++++++++++++++++++++++++-------
>> lib/tools/emacs/test.erl.indented | 38 ++++++++++++++++++++++++
>> lib/tools/emacs/test.erl.orig | 38 ++++++++++++++++++++++++
>> 3 files changed, 122 insertions(+), 12 deletions(-)
>>
>> diff --git a/lib/tools/emacs/erlang.el b/lib/tools/emacs/erlang.el
>> index bc7a190..2f6c7f5 100644
>> --- a/lib/tools/emacs/erlang.el
>> +++ b/lib/tools/emacs/erlang.el
>> @@ -2986,18 +2986,52 @@ This assumes that the preceding expression is either simple
>> (forward-sexp (- arg))
>> (let ((col (current-column)))
>> (skip-chars-backward " \t")
>> - ;; Needed to match the colon in "'foo':'bar'".
>> - (if (not (memq (preceding-char) '(?# ?:)))
>> - col
>> - ;; Special hack to handle: (note line break)
>> - ;; [#myrecord{
>> - ;; foo = foo}]
>> - (or
>> - (ignore-errors
>> - (backward-char 1)
>> - (forward-sexp -1)
>> - (current-column))
>> - col)))))
>> + ;; Special hack to handle: (note line break)
>> + ;; [#myrecord{
>> + ;; foo = foo}]
>> + ;; where the call (forward-sexp -1) will fail when point is at the `#'.
>> + (or
>> + (ignore-errors
>> + ;; Needed to match the colon in "'foo':'bar'".
>> + (cond ((eq (preceding-char) ?:)
>> + (backward-char 1)
>> + (forward-sexp -1)
>> + (current-column))
>> + ((eq (preceding-char) ?#)
>> + ;; We may now be at:
>> + ;; - either a construction of a new record
>> + ;; - or update of a record, in which case we want
>> + ;; the column of the expression to be updated.
>> + ;;
>> + ;; To see which of the two cases we are at, we first
>> + ;; move an expression backwards, check for keywords,
>> + ;; then immediately an expression forwards. Moving
>> + ;; backwards skips past tokens like `,' or `->', but
>> + ;; when moving forwards again, we won't skip past such
>> + ;; tokens. We use this: if, after having moved
>> + ;; forwards, we're back where we started, then it was
>> + ;; a record update.
>> + ;; The check for keywords is to detect cases like:
>> + ;; case Something of #record_construction{...}
>> + (backward-char 1)
>> + (let ((record-start (point))
>> + (record-start-col (current-column)))
>> + (forward-sexp -1)
>> + (let ((preceding-expr-col (current-column))
>> + ;; white space definition according to erl_scan
>> + (white-space "\000-\040\200-\240"))
>> + (if (erlang-at-keyword)
>> + ;; The (forward-sexp -1) call moved past a keyword
>> + (1+ record-start-col)
>> + (forward-sexp 1)
>> + (skip-chars-forward white-space record-start)
>> + ;; Are we back where we started? If so, it was an update.
>> + (if (= (point) record-start)
>> + preceding-expr-col
>> + (goto-char record-start)
>> + (1+ (current-column)))))))
>> + (t col)))
>> + col))))
>>
>> (defun erlang-indent-parenthesis (stack-position)
>> (let ((previous (erlang-indent-find-preceding-expr)))
>> diff --git a/lib/tools/emacs/test.erl.indented b/lib/tools/emacs/test.erl.indented
>> index 2948ccf..e0593c6 100644
>> --- a/lib/tools/emacs/test.erl.indented
>> +++ b/lib/tools/emacs/test.erl.indented
>> @@ -657,3 +657,41 @@ indent_comprehensions() ->
>> foo() ->
>> [#foo{
>> foo = foo}].
>> +
>> +%% Record indentation
>> +some_function_with_a_very_long_name() ->
>> + #'a-long-record-name-like-it-sometimes-is-with-asn.1-records'{
>> + field1=a,
>> + field2=b},
>> + case dummy_function_with_a_very_very_long_name(x) of
>> + #'a-long-record-name-like-it-sometimes-is-with-asn.1-records'{
>> + field1=a,
>> + field2=b} ->
>> + ok;
>> + Var = #'a-long-record-name-like-it-sometimes-is-with-asn.1-records'{
>> + field1=a,
>> + field2=b} ->
>> + Var#'a-long-record-name-like-it-sometimes-is-with-asn.1-records'{
>> + field1=a,
>> + field2=b};
>> + #xyz{
>> + a=1,
>> + b=2} ->
>> + ok
>> + end.
>> +
>> +another_function_with_a_very_very_long_name() ->
>> + #rec{
>> + field1=1,
>> + field2=1}.
>> +
>> +some_function_name_xyz(xyzzy, #some_record{
>> + field1=Field1,
>> + field2=Field2}) ->
>> + SomeVariable = f(#'Some-long-record-name'{
>> + field_a = 1,
>> + 'inter-xyz-parameters' =
>> + #'Some-other-very-long-record-name'{
>> + field2 = Field1,
>> + field2 = Field2}}),
>> + {ok, SomeVariable}.
>> diff --git a/lib/tools/emacs/test.erl.orig b/lib/tools/emacs/test.erl.orig
>> index 1221c56..69356ac 100644
>> --- a/lib/tools/emacs/test.erl.orig
>> +++ b/lib/tools/emacs/test.erl.orig
>> @@ -657,3 +657,41 @@ ok.
>> foo() ->
>> [#foo{
>> foo = foo}].
>> +
>> +%% Record indentation
>> +some_function_with_a_very_long_name() ->
>> + #'a-long-record-name-like-it-sometimes-is-with-asn.1-records'{
>> + field1=a,
>> + field2=b},
>> + case dummy_function_with_a_very_very_long_name(x) of
>> + #'a-long-record-name-like-it-sometimes-is-with-asn.1-records'{
>> + field1=a,
>> + field2=b} ->
>> + ok;
>> + Var = #'a-long-record-name-like-it-sometimes-is-with-asn.1-records'{
>> + field1=a,
>> + field2=b} ->
>> + Var#'a-long-record-name-like-it-sometimes-is-with-asn.1-records'{
>> + field1=a,
>> + field2=b};
>> + #xyz{
>> + a=1,
>> + b=2} ->
>> + ok
>> + end.
>> +
>> +another_function_with_a_very_very_long_name() ->
>> + #rec{
>> + field1=1,
>> + field2=1}.
>> +
>> +some_function_name_xyz(xyzzy, #some_record{
>> + field1=Field1,
>> + field2=Field2}) ->
>> + SomeVariable = f(#'Some-long-record-name'{
>> + field_a = 1,
>> + 'inter-xyz-parameters' =
>> + #'Some-other-very-long-record-name'{
>> + field2 = Field1,
>> + field2 = Field2}}),
>> + {ok, SomeVariable}.
>> --
>> 1.7.5.4
>>
> _______________________________________________
> erlang-patches mailing list
> erlang-patches@REDACTED
> http://erlang.org/mailman/listinfo/erlang-patches
--
/Henrik Nord Erlang/OTP
More information about the erlang-patches
mailing list