Re: [PATCHES] various commenting, whitespace changes

[Thien-Thi Nguyen <ttn@gnuvola.org>]

() Andy Wingo <wingo@pobox.com>
() Mon, 28 Jun 2010 15:44:54 +0200

Sorry for the long delay in response.

   What does NFC mean?

See LibreDWG HACKING (and build-aux/create-changelog), for more info:
git://git.sv.gnu.org/libredwg.git

It's useful to distinguish between "important" and "unimportant"
changes, especially if recording changes (ChangeLog maintenance) is
mechanical.  A simple tag like this gives you the flexibility to make
unimportant changes without the worry that such will end up as noise in
the published ChangeLog.

   In general I'm a little uneasy with this kind of change, as it is
   unnecessary.

No worries, lots of aesthetic/readability munging that i do falls into
that category -- you'll get used to it.  I think if you cut me some
slack on this matter (and perhaps implement "nfc" (or similar) filtering
as per LibreDWG create-changelog), you may find yourself not only more
accepting of these kinds of changes but also doing them yourself.

   But, OK, I guess. Please feel free to make files more closely follow
   GNU conventions, when you are working on those files; but do avoid a
   general cleanup. I think it would drive me crazy ;) Perhaps Ludovic
   will chime in later with more general feedback.

I will avoid general cleanup, but i don't intend to avoid cleanup on
files i wish to hack on (for now, this set includes makefiles, build
scripts and documentation).

I hope this targeted approach will not drive you crazy.  Rather, it
would be ideal if you were to say "hey, thanks janitor ttn for handling
that grunge -- keep up the good work!", after observing that these
changes can coincide peacefully with your own hacking sphere (if you let
that sphere become more transparent :-).

   > --- a/libguile/guile-func-name-check
   > +++ b/libguile/guile-func-name-check
   [...]
   >  BEGIN {
   > -  filename = ARGV[1];
   > -  in_a_func = 0;
   > +    filename = ARGV[1];
   > +    in_a_func = 0;
   >  }

   There is no need to change the indentation amount here; 2-space should
   be fine.

   [...]
   > +# guile-func-name-check ends here

   I would rather not have these markers. It's an extra point of bitrot.

   If you really want the end-of-file markers I would want some
   confirmation from Ludovic; otherwise, I am OK with you pushing these
   patches without the EOF markers and without changing the awk script
   indentation.

I sidestepped these issues, by rewriting in Scheme (and using Emacs'
‘indent-sexp’).  Unfortunately, it looks like hydra is not liking it:

http://hydra.nixos.org/build/606792

Perhaps a workaround would be to use ‘--no-autocompile’ in the
invocation command (libguile/guile-snarf-docs.in:55), but that seems
ugly and throws away a bug-{hunt,squash}ing opportunity.  WDYT?

Another related idea: Since Guile supports many languages, why not AWK?