Re: [Nasm-devel] Indenting the source

[Ed B. <be...@mi...>]

H. Peter Anvin wrote:
> Ed Beroset wrote:
> 
>>
>> Therefore, while HPA's line specifies more (which is a good thing, 
>> IMHO) the only actual *differences* are as follows:
>>
>> Ed: -ci8 -nlp
>> HPA: -ci4 -lp
>>
>> To borrow an example from the indent man pages, the difference would 
>> look like this with -ci8 -nlp:
>>
>> p1 = first_procedure (second_procedure (p2, p3),
>>         third_procedure (p4, p5));
>>
>> With -ci4 -lp in effect the code looks like this:
>>
>> p1 = first_procedure (second_procedure (p2, p3),
>>                       third_procedure (p4, p5));
>>
>>
>> Because these are minor differences, I could endorse either, but I 
>> would very much like to have one consistently applied style 
>> throughout, no matter what that style.
>>
> 
> OK, both of those I feel fairly strongly about.  I really want the 
> difference between statements like if, for, and while and function 
> calls, and NASM, like most compilers, is really way too nested to be 
> well handled by an indentation level of 8 and still have space for 
> comments.
> 
> Linus doesn't believe in comments, so he advocates indentation by 8.  To 
> quote Andrew Morton, "however, I do."

I think you may misunderstand the difference.  The -ci8 is just the 
*continuation* indent, meaning a single long line of code is wrapped and 
indented by 8.  The -i4 means that the normal indentation is still 4, as 
we both have indicated we prefer.  Most of the code seems to be in that 
style already.

Also, by preferring -lp instead of -nlp, you actually leave less room 
for comments, not more.  The example statements above illustrate both of 
these points.  The -lp version has the function arguments lined up 
vertically, and the -nlp -ci8 just indents the second line by 8 spaces 
instead as many as it takes to line up the parameters.

> If this is OK with everyone else, I'll run indent over the sources and 
> check those in as a separate checkin.  It's still a flag day -- you 
> effectively lose all change tracking across the reformat -- but at least 
> there are no other changes involved.

Yes, I'd strongly endorse this, but only if we agree first on which 
style to use.  As you say, it's a flag day, and so we'd better only do 
it once!

Ed