On Wednesday 04 July 2001  7:08 pm, Greg A. Woods wrote:
> [ On Wednesday, July 4, 2001 at 09:30:23 (+0100), Chris Gilbert wrote: ]
>
> > Subject: Re: build of -current from today fails on libexec/kpasswdd
> >
> > kdump and makewhatis appear to need DESTDIR in the env,
>
> No, they don't.  Where did you get the idea that they do?  They don't
> even have DESTDIR anywhere near them, let alone inside them!
>
> (and where is kdump used during the build?  perhaps you did an fgrep for
> DESTDIR in the source and didn't read the context to see what the actual
> use was for?  it's only in a makefile fragment called "Makefile.ioctl-c")

Yes I did an fgrep, ignoring any file with Makefile in it's name and found 
that kdump/mkioctls uses DESTDIR, and that's a shell script called by 
Makefile.ioctl-c, which is included by the Makefile. (unless I'm also not 
reading the Makefile's closely enough again)

makewhatis I'll willingly admit to being wrong on, I did a grep and noted 
that src/libexec/makewhatis/makewhatis.sh has references to DESTDIR, however 
closer examination reveals that the .sh file is now obsoleted by the c 
version, perhaps we should actually remove it from cvs now?

> > some arch's
> > floppy/ram disk image creation scripts need it as well.
>
> If they do then they are broken and need fixing.  The fix is extremely
> trivial and obvious.  There shouldn't be any such problems on i386 or
> sparc, IIRC, though I have re-written many of those makefiles myself and
> all I really know is that my new versions have no such problems.

Yes they may be broken, my point is that currently they use it and as such 
need DESTDIR in the env.  (as I did say this may not be the desired way, but 
it's how things currently stands)

> > I got bitten by this when building 1.5.x while running -current, and also
> > when building -current when running 1.5.x as kdump gets confused :)
>
> I've just recently build release snapshots from "-current" (from
> 2001/06/24) and I've encountered no such problems.  I *never* specify
> build-related variables in the environment.

fair enough, I just know I had some issue with the process somewhere, the 
rectified itself with setting DESTDIR, at the time I was trying to sort out a 
snapshot, certainly arm32 disk images look to need DESTDIR so that might have 
been why.

I agree that we shouldn't need to set DESTDIR in the env, but currently I've 
found I need to.

Cheers,
Chris