[Pkg-pciutils-discuss] Time for a new release

[Matthew Wilcox]

On Fri, Aug 18, 2006 at 02:39:07AM -0700, Matt Taggart wrote:
> I just uploaded pciutils 2.2.4~pre4-1 (and tagged in cvs)
> I also uploaded libpci2 2.1.11-3

Thanks!

> > We've talked about splitting out pci.ids into its own package,
> 
> Should this wait until after etch? The only reason I can think of for doing 
> this before etch is that it might make updating in stable releases easier,
> but doing a new pciutils with only pci.ids updated doesn't seem any harder.

Oh, it would definitely be easier; it'd be arch:all instead of arch:any.

> > But I don't think
> > this is the last upload before etch, so there's still time for someone to
> > make those changes after this release.
> 
> We are quickly approaching our last opportunity though.

Yes, and I hear that since we provide a udeb, we're already frozen.  But
we're not on the testing hints freeze list, so I dunno what's really
going on.

> Martin,
> It might be nice if we could get the debian diffs minimized and if you did a 
> final update to pci.ids and then released 2.2.4, as this will probably be the 
> version to go in etch and live for a while.
> 
> Here is a link to the Debian diff for the version I just uploaded,
> 
>  http://incoming.debian.org/pciutils_2.2.4~pre4-1.diff.gz

It won't be there for long ... after today it'll be at
http://ftp.debian.org/pool/main/p/pciutils/pciutils_2.2.4~pre4-1.diff.gz

> Other than the debian/ stuff (which shouldn't go upstream), the pcimodules  
> stuff (which willy is trying to get rid of), and the pci.ids diff there's not 
> much left. Can any of that go upstream?

I suspect it's worth submitting a patch to add DESTDIR support.
It's intertwined with pcimodules support right now, which is why I
haven't been bothered to tackle it yet.

We're still trying to find a justification for the change to lspci.c.
I don't think anyone was able to come up with a reason why we shouldn't
be able to read the first 64 bytes from a device.

I don't know if Martin would be interested in compiling with the -g flag
by default.  I know he's not interested in the -fPIC option and wants
us to keep that as part of the Debian diff.

I haven't asked Martin about the change to example.c to use "pci/pci.h".
I suspect it's there to allow example.c to be compiled against
pciutils-dev once installed rather than having to be compiled as part
of pciutils.

I have more in my development tree which I'd like to have merged.
The support for gzipped pci.ids and the byte->char conversion to remove
gcc warnings.