Re: Patch for dpkg-scansources / Packages without sources
Raphael Hertzog wrote:
> --- dpkg-scansources.old Thu Jul 29 12:00:54 1999
> +++ dpkg-scansources Thu Jul 29 12:00:57 1999
> @@ -232,7 +232,7 @@
>
> sub de_pgp {
> my ($file, $s) = @_;
> - if ($s =~ s/^-----BEGIN PGP SIGNED MESSAGE-----\n\n//) {
> + if ($s =~ s/^-----BEGIN PGP SIGNED MESSAGE-----\n(?:Hash: \S+\n)?\n//) {
> unless ($s =~ s/\n
> -----BEGIN\040PGP\040SIGNATURE-----\n
> .*?\n
This is not quite correct. Instead of looking specifically for Hash:,
you should skip until after the first empty line. PGP headers
function just like "normal" headers in this regard. (The only
difference is that normally there are zero header lines).
Sorry, it's too early in the morning for safe perling for me.
> Furthermore, there are many packages that are no more built by any source
> packages. Many should be removed, in general they have been renamed or
> splitted, here's the list (it's does come from a log from my CD build
> script) :
>
> I wonder why libstdc++2.9 is no more built by any sources ... this one
> shouldn't be removed. :) Also some kernel related packages do not have
> "official" sources (ie the packages are not mentionned in a "Binary"
> field in the Sources.gz files).
>
> Package `liblockfile0' has no sources ...
Superseded by liblockfile1, so it will go.
> Package `libstdc++2.9' has no sources ...
> Package `libg++2.8.2' has no sources ...
> Package `libg++2.8.2-dev' has no sources ...
The current packages are libstdc++2.9-glibc2.1 and
libg++2.8.2-glibc2.1. libstdc++2.9 and libg++2.8.2 will have to go
before the release, because we can't build them in potato anymore.
> Package `lsof-2.2' has no sources ...
> Package `lsof-2.0.35' has no sources ...
I think lsof-2.0.35 should be removed, but lsof-2.2 is probably just
built strangely.
> Package `ftape-module-2.0.35' has no sources ...
This is built strangely, like lsof and pcmcia-modules.
> Package `pcmcia-modules-2.0.35' has no sources ...
> Package `pcmcia-modules-2.0.36' has no sources ...
> Package `pcmcia-modules-2.2.1' has no sources ...
> Package `pcmcia-modules-2.2.10' has no sources ...
> Package `pcmcia-modules-2.2.5' has no sources ...
> Package `pcmcia-modules-2.2.7' has no sources ...
> Package `pcmcia-modules-2.2.9' has no sources ...
See above. Probably some of them can go. I wish these packages were
built like Wichert's kernel-module packages.
> Package `c-client-dev' has no sources ...
Hmm... I remember accepting a c-client package recently. Maybe it
got renamed to libc-client-dev or something.
> Package `clanlib0-dev' has no sources ...
Not sure about this one.
> Package `libcdaudio-dev' has no sources ...
> Package `libcdaudio' has no sources ...
Superseded by libcdaudio0. These will go.
> Package `libgimp1.1.4-dev' has no sources ...
> Package `libgimp1.1.5-dev' has no sources ...
> Package `libgimp1.1.6-dev' has no sources ...
> Package `libgimp1.1.6' has no sources ...
> Package `libgimp1.1.4' has no sources ...
> Package `libgimp1.1.5' has no sources ...
More old libraries. These change soname frequently.
> Package `libwraster2-dev' has no sources ...
> Package `libwraster2' has no sources ...
Ah, the tale of libwraster... Right now, the newest version is
*probably* libwraster1 :-)
> Package `libcompface1' has no sources ...
> Package `libcompface1-altdev' has no sources ...
These are bogus. libcompface was NMUd on a platform which has no
libc5, so the libc5-compat packages don't appear in its source's
Binary line anymore. I guess libcompface has no maintainer anymore,
since it's been like this for months.
> Package `libxdelta1' has no sources ...
> Package `libxdelta1-dev' has no sources ...
Superseded by libxdelta2, will be removed.
> Package `libpwdb-doc' has no sources ...
> Package `mysql-gpl-dev' has no sources ...
> Package `popt' has no sources ...
> Package `gnat-rm' has no sources ...
> Package `gnat-user-guide' has no sources ...
> Package `kernel-doc-2.1.125' has no sources ...
> Package `kbd-data' has no sources ...
> Package `xcin-doc' has no sources ...
These will require further investigation. (I'm offline at the moment).
We have scripts for detecting this sort of thing, but I don't often have
time to run them and fix things. I think your scripts will have to be
able to live with it, because I doubt the archive will ever be completely
happy about source-binary package relationships. You should see how much
bogus output the scripts generate... there's the pcmcia stuff, the libc6.1
madness, the egcs confusion, the libcompface situation, and lots more fun
for the whole family.
Richard Braakman
Reply to: