Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 11 Feb 2018 18:29:10 -0700
From:      Warner Losh <imp@bsdimp.com>
To:        Bruce Evans <brde@optusnet.com.au>
Cc:        Warner Losh <imp@freebsd.org>, src-committers <src-committers@freebsd.org>, svn-src-all@freebsd.org, svn-src-head@freebsd.org
Subject:   Re: svn commit: r329127 - head/sys/sys
Message-ID:  <CANCZdfp=9KtG_Ez3o0dK-sZyoCevS91ELab0Kb996vbCPEpcYg@mail.gmail.com>
In-Reply-To: <20180212111457.R873@besplex.bde.org>
References:  <201802111745.w1BHjcWA050692@repo.freebsd.org> <20180212111457.R873@besplex.bde.org>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sun, Feb 11, 2018 at 5:55 PM, Bruce Evans <brde@optusnet.com.au> wrote:

> On Sun, 11 Feb 2018, Warner Losh wrote:
>
> Log:
>>  Consistent macro indentation is the hobgoblin of small minds
>>
>>  Line up the macro definitions and names here like the machine/stdarg.h
>>  files that this replaced. This is easier to read and also makes it
>>  easier to match up with other includes. Also two space indent va_end
>>  to match the rest of the surrounding if block.
>>
>
> Any chance of using KNF style?
>
> The 2-column indent for cpp things is easier to read than the KNF 0-column
> indent for cpp things, but is inconsistent with the KNF 8-column indent for
> C things, and is harder to write and maintain.
>
> Modified: head/sys/sys/_stdarg.h
>> ============================================================
>> ==================
>> --- head/sys/sys/_stdarg.h      Sun Feb 11 16:35:56 2018        (r329126)
>> +++ head/sys/sys/_stdarg.h      Sun Feb 11 17:45:38 2018        (r329127)
>> @@ -43,13 +43,13 @@
>> #endif
>>
>> #ifdef __GNUCLIKE_BUILTIN_STDARG
>> -  #define va_start(ap, last) __builtin_va_start((ap), (last))
>> -  #define va_arg(ap, type) __builtin_va_arg((ap), type)
>> -  #define __va_copy(dest, src) __builtin_va_copy((dest), (src))
>> +  #define      va_start(ap, last)      __builtin_va_start((ap), (last))
>> +  #define      va_arg(ap, type)        __builtin_va_arg((ap), type)
>> +  #define      __va_copy(dest, src)    __builtin_va_copy((dest), (src))
>>
>
> In KNF, '#define' is not indented, and the tab after '#define' puts the
> name in column 8.  Here, '#define is indented by 2 columns, and the space
> after #define used to put the name in column 2+7+1 = 10; now it puts the
> name in column 16 which is too far to the right.
>
>   #if __ISO_C_VISIBLE >= 1999
>> -    #define va_copy(dest, src) __va_copy(dest, src)
>> +    #define    va_copy(dest, src)      __va_copy(dest, src)
>>
>
> Deeper nesting causes further problems.  Now the '#define's are indented
> by 2+2 columns.  The names used to be indented to match, but they are
> now indented uniformly to 16 (except for unconverted ones), so their
> indentation no longer reflects the nesting.  For 5 levels of nesting,
> the #define's would be in column 10 and the tab would indent the names
> to column 24.
>
>   #endif
>> -#define va_end(ap) __builtin_va_end(ap)
>> +  #define      va_end(ap)              __builtin_va_end(ap)
>>
>
> This like was actually in KNF style.


Right, but it didn't match the rest of the file :)


>
> #endif
>>
>> #if defined(lint) && !defined(va_start)
>>
>
> There are many unchanged misformattings before and after the ones modified
> in the patch.  These are now more inconsistent than before:
> before:
> - space instead of tab after #define for idempotency ifdef
> - space instead of tab after #define for _VA_LIST_DECLARED
> after:
> - no tabs in the lint section (4 #define's).  One of these is long and
>   would blow out to 92 columns with tabs instead of spaces.  After fixing
>   the indentation of #define, it still takes 83 columns.
>
> Other style bugs in this file:
> - _VA_LIST_DECLARED is declared before va_list is actually declared.  This
>   is illogical, and its bad style is not used for older #define's of this
>   type
> - space after ! in comment on #endif for idempotency ifdef
> - garbage newline before EOF.
>
> This file scores 67% on knfom(1) after this commit, down from 68% before.
> Most of the errors detected are for the 2-column indentation of almost
> everything.  After fixing this, the score is 95%.  IIRC, indent(1) never
> reformats cpp directives.  knfom(1) uses indent(1) so it doesn't notice
> bugs in cpp expressions.  The score is only 95% because of bugs in
> indent(1).  indent(1) fixes the cpp indentation and the garbage before EOF,
> but adds the following bugs:
> - change the tab in tyhe typedef to a space
> - indent the comment on the #endif for the idempotency ifdef.


what's knfom? Google gives me garbage when I search for that.

Warner



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CANCZdfp=9KtG_Ez3o0dK-sZyoCevS91ELab0Kb996vbCPEpcYg>