[Nut-upsdev] Fwd: m4 scripts ignore user-supplied CFLAGS and LDFLAGS

[Peter Selinger]

Hi Alexander,

no, I am not responsible for this! It's true that I created the
scripts in m4/, but I only copied what was previously in configure.in.
Try:

svn up -r2 configure.in

and you'll see that the code in question has been there since the
beginning of SVN time (line 175).

I cannot say why it was done that way. Probably because nobody needed
a more general mechanism. Using pkgconfig seems like a good idea, but
there should be a fallback method if pkgconfig is not installed or
does not work. I think some of our autoconf tests already do something
like this.

-- Peter

Alexander I. Gordeev wrote:
> 
> Hi Peter!
> 
> Sorry for disturbing you...
> I've checked svn logs for m4 macros and it seems that you
> are the only person to ask :)
> 
> ------- Forwarded message -------
> From: "Alexander I. Gordeev" <lasaine at lvk.cs.msu.su>
> To: nut-upsdev at lists.alioth.debian.org
> Cc:
> Subject: [Nut-upsdev] m4 scripts ignore user-supplied CFLAGS and LDFLAGS=
> 
> Date: Thu, 21 Feb 2008 03:33:26 +0300
> 
> Hi all,
> 
> There is a discussion going on one russian forum about NUT
> cross-compiling and packaging for some custom embedded firmware.
> And one of the authors complained about NUT script for checking
> ssl library ignoring supplied CFLAGS and LDFLAGS with the
> desired include and library paths.
> I checked several scripts in m4 folder and all of them do
> something like this:
> 
>      CFLAGS_ORIG=3D"${CFLAGS}"
>      LDFLAGS_ORIG=3D"${LDFLAGS}"
> 
>      CFLAGS=3D""
>      LDFLAGS=3D"..."
> 
> at the beginning. Is it for purpose? Why not check first with
> the user-supplied variables?
> 
> I'm not very good in all this autotools magic so please
> forgive me if this whole thing is dumb :)
> 
> P.S. What is the "right" way of passing custom paths (for a
> cross-compilation environment)?
> 
> ------- Forwarded message -------
> 
> P.S. I think we should consider pkg-config, right?
> 
> -- =
> 
>    Alexander
>