Re: [Ltib] package preconfig check confusion

[Stuart Hughes]

On 25/09/13 02:33, Mike Goins wrote:
> On Tue, Sep 24, 2013 at 9:14 PM, Mike Goins <address@hidden> wrote:
>> I'm having a bit of an issue with understanding a bit of logic with
>> preconfig and dir builds.
>>
>> A kernel is under dir build.
>> Kernel preconfig gets updated through version control.
>> ltib runs and outputs this:
>>
>> Processing: kernel-3.9.11-mpc8377
>> ============================================
>> Build path taken because: preconfig linux-3.9.11-mpc8377.config newer
>> than rpm, directory build,
>> checking if sources have been updated: no
>> ...
>> goes to rest of packages without building the kernel.
>>
>>
>> It looks as if ltib detected that the preconfig is newer, but doesn't
>> force the re-build of the package, like dir build takes precedence
>> over preconfig.   I don't have PKG_KERNEL_FORCE selected, and thought
>> it a bit heavy handed.  Shouldn't the newer preconfig auto-trigger a
>> force rebuild?
>>
>> It does appear that setting PKG_KERNEL_FORCE, would do it, but this is
>> such an expensive operation, when 99% of the time it is not needed
>> (I'd rather not trigger the kernel build counter unless something
>> actually changed).
> 
> 
> Possibly answering my own question:
> 
> Index: ltib
> ===================================================================
> RCS file: /sources/ltib/ltib/ltib,v
> retrieving revision 1.89
> diff -u -r1.89 ltib
> --- ltib 20 Oct 2012 14:05:42 -0000 1.89
> +++ ltib 25 Sep 2013 01:30:53 -0000
> @@ -831,7 +831,7 @@
>          # don't do this clause if running in short-circuited mode
>          if( $unpack eq 'yes' )  {
>              # commit to installing a new rpm (enforced if build fails)
> -            unlink(@rpms) if $cf->{force} || $$key->{build} || $spec_upd;
> +            unlink(@rpms) if $cf->{force} || $$key->{build} ||
> $spec_upd || $preconfig_bld;
> 
>              foreach my $url (  split(/\s*\n/, $tok->{sources}),
>                                 split(/\s*\n/, $tok->{patches})   ) {
> 

Hi Mike,

I took a look at the code and tried out your patch.  It looks good to
me, please check-in.

Regards, Stuart