On Wednesday 19 September 2007, Stepan Kasal wrote:
> Hi,
>
> On Wed, Sep 19, 2007 at 10:32:13AM -0400, Mike Frysinger wrote:
> > i figured it was to make things like vim happy ... perhaps it should be:
> > dnl #(
> > so that the #( isnt output to the final file ...
>
> well, in that case you could omit the #, using "dnl (".
>
> The tradition in Autoconf sources is to use #( but I have to admit
> that I'm not sure about the rationale.
>
> Possible reasons:
> - #( is shorter and easier to distinguish visually,
> - when debugging/inspecting the generated shell code, matched parens
> might help,
> - "dnl (" does not work inside quoted (literal) shell code, so you
> have to be more careful.
>
> None of these really matters but, OTOH, I don't buy your reason
> either:
>
> Having #( is not an issue, the effect on size, speed, nor readability
> is not noticable.
>
> (If you are intersted, feel free to bring this question to the
> Autoconf list.)
if you read enough .m4's you'll see people tend to be very anal about spurious
newlines/whitespace as the logic "one isnt a lot" doesnt hold up as
everything quickly snowballs
that's why there are so many lines like:
[dnl
...
])dnl
spread throughout all .m4's from any of the autotools packages
here, it adds little to the generated configure since we use the macro too
often ... i was just curious as to why it was there at all
-mike
Attachment:
signature.asc
Description: This is a digitally signed message part.
