mode for preprocessed files

mode for preprocessed files

[Dan Nicolaescu]

Currently C and C++ preprocessed files (.i and .ii, respectively) are
opened in fundamental mode.

Would anyone object if we would use c-mode and c++-mode for them by default?
I've been doing this locally for a long time without any problem.

Re: mode for preprocessed files

[Stefan Monnier]

> Currently C and C++ preprocessed files (.i and .ii, respectively) are
> opened in fundamental mode.

> Would anyone object if we would use c-mode and c++-mode for them by default?
> I've been doing this locally for a long time without any problem.

It might be OK, tho I wonder how often that's useful.
I.e. it's OK but put it way down the list, with a comment explaining why
.i and .ii are mapped to c-mode and c++-mode (it wouldn't have occurred
to me that those extensions are actually used for C and C++ files).


        Stefan

Re: mode for preprocessed files

[David De La Harpe Golden]

Dan Nicolaescu wrote:
> Currently C and C++ preprocessed files (.i and .ii, respectively) are
> opened in fundamental mode.
>

Well, it's what gcc -save-temps uses...

> Would anyone object if we would use c-mode and c++-mode for them by default?
> I've been doing this locally for a long time without any problem.
> 

FWIW, I still use .i, just out of habit, for asm includes, though
I guess that's not widespread anymore (was standard on Amiga), and I
should probably stop or one day I'll gcc -save-temps something important
into oblivion...

Re: mode for preprocessed files

[Dan Nicolaescu]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

  > > Currently C and C++ preprocessed files (.i and .ii, respectively) are
  > > opened in fundamental mode.
  > 
  > > Would anyone object if we would use c-mode and c++-mode for them by default?
  > > I've been doing this locally for a long time without any problem.
  > 
  > It might be OK, tho I wonder how often that's useful.
  > I.e. it's OK but put it way down the list, with a comment explaining why
  > .i and .ii are mapped to c-mode and c++-mode

Done.  But the list is created using autoloads... BTW, wasn't there a
plan to avoid doing that?

  >  (it wouldn't have occurred
  > to me that those extensions are actually used for C and C++ files).

They are the standard extensions that C/C++ compilers generate when you
tell them to save the preprocessed file.  Compilers accept files with
those extensions without wondering what the language is.

Re: mode for preprocessed files

[Stefan Monnier]

>> (it wouldn't have occurred
>> to me that those extensions are actually used for C and C++ files).
> They are the standard extensions that C/C++ compilers generate when you
> tell them to save the preprocessed file.

I know, but in all my years of C coding, I've never seen this
"feature" used.  I've seen things piped through cpp, but not saved to
a .i(i) file.

> Compilers accept files with
> those extensions without wondering what the language is.

Of course, they don't need to wonder if it might be something completely
different since the mere fact of passing such a file to the compiler
already tells the compiler that it's some kind of source file, so at
worst the compiler has to decide which of the compiler's supported
source languages is used.  Emacs has to take into account whether such
extensions are used in completely different contexts.


        Stefan


PS: the .i and .ii extensions are not in Debian's /etc/mime.types, for example.

Re: mode for preprocessed files

[Dan Nicolaescu]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

  > >> (it wouldn't have occurred
  > >> to me that those extensions are actually used for C and C++ files).
  > > They are the standard extensions that C/C++ compilers generate when you
  > > tell them to save the preprocessed file.
  > 
  > I know, but in all my years of C coding, I've never seen this
  > "feature" used.  I've seen things piped through cpp, but not saved to
  > a .i(i) file.

When working on the compiler preprocessed files are pretty much the
norm.  Yes, compiler people are a small minority, but they make the
world go around ... :-)

Re: mode for preprocessed files

[Miles Bader]

Stefan Monnier <monnier@iro.umontreal.ca> writes:
> I know, but in all my years of C coding, I've never seen this
> "feature" used.  I've seen things piped through cpp, but not saved to
> a .i(i) file.

Hmm, I've certainly run across .i for such, and indeed, I seem to use
that extension for my own preprocessed files in a few of my makefiles
where I have rules for making them.

No idea where I picked that up though.

-Miles

-- 
((lambda (x) (list x x)) (lambda (x) (list x x)))

Re: mode for preprocessed files

[Richard M Stallman]

I have occasionally saved CPP output in files for the purpose of
debugging GCC.  I don't think I ever used the extension .i for it, but
if there is such a convention now I suppose I would follow it.