[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Packages in Auto-Not-For-Us & Failed (etc)



* Philipp Kern (pkern@debian.org) [140412 20:26]:
> On 2014-04-12 18:01, Andreas Barth wrote:
>> currently packages which should get into the state Auto-Not-For-Us
>> (e.g. arch not in arch list) and are in one of the states: Not-For-Us
>> Failed Failed-Removed Dep-Wait Dep-Wait-Removed are not touched.
>>
>> This is wrong for two reasons: The substate uncompiled/out-of-date is
>> not updated. The other is that it might make sense to move the package
>> to Auto-Not-For-Us anyways, because there was a deliberate action for
>> that. What do you think?
>
> For the *-Removed ones I'd say that we should just --forget them. If  
> they reappear as auto-nfu, that's ok.

The *-Removed are more or less historic-only and are mostly handleded
as s,-Removed,,.

> If you know that a package would otherwise land in auto-nfu, I would not  
> mind if you clean up the database and get them into auto-nfu.

I think about this change to the code:
--- a/bin/wanna-build
+++ b/bin/wanna-build
@@ -2323,7 +2323,7 @@ sub parse_all_v3 {
             my $tstate = {'installed' => 'Installed', 'related' => 'Installed', 
                 'arch-not-in-arch-list' => 'Auto-Not-For-Us', 'packages-arch-specific' => 'Auto-Not-For-Us', 'overwritten-by-arch-all' => 'Auto-Not-For-Us', 'arch-all-only' => 'Auto-Not-For-Us',
                 }->{$pkgs->{'status'}};
-            next if isin( $pkg->{'state'}, qw<Not-For-Us Failed Failed-Removed Dep-Wait Dep-Wait-Removed>) && isin( $tstate, qw<Auto-Not-For-Us>);
+            next if isin( $pkg->{'state'}, qw<Not-For-Us>) && isin( $tstate, qw<Auto-Not-For-Us>);
             # if the package is currently current, the status is Installed, not not-for-us
             if ($pkg->{'state'} ne $tstate) {
                 change_state( \$pkg, $tstate);


Which would just automaticall push packages from the states Failed
Failed-Removed Dep-Wait Dep-Wait-Removed to auto-nfu in case that
would be the default answer, and keep n-f-u as-is. (One could extend
that to all of them, or whatever states should be used; the *-Removed
are mostly or fully history, and should perhaps just be dropped. So no
hand-cleanup but code-cleanup.)


Andi


Reply to: