On 27/05/25 at 00:59 +0200, Santiago Vila wrote:
> > Could these be selected with UDD for all uploads between two dates
> > (then and now).
>
> Ok, here is a list of packages meeting the following conditions:
>
> - build-depends on cdbs
> - last uploaded between 2025-04-08 and 2025-05-27
>
> 2025-04-09 libiptcdata 1.0.5-2.4
> 2025-04-09 vim-scripts 20210124.4
> 2025-04-13 oss4 4.2-build2020-5
> 2025-04-24 x2gobroker 0.0.4.3-4.3
> 2025-05-04 dhelp 0.6.31
> 2025-05-04 libmessage-passing-filter-regexp-perl 0.05-5
> 2025-05-04 libmodule-install-copyright-perl 0.009-4
> 2025-05-04 libmodule-install-doapchangesets-perl 0.206-4
> 2025-05-04 libmodule-install-doap-perl 0.006-4
> 2025-05-04 libmodule-install-rdf-perl 0.009-4
> 2025-05-04 libmoosex-arrayref-perl 0.005-4
> 2025-05-04 libpath-router-perl 0.15-4
> 2025-05-04 librdf-trinex-serializer-mockturtlesoup-perl 0.006-4
> 2025-05-04 librdf-vcard-perl 0.012-4
> 2025-05-05 haskell-ogma-core 1.7.0-2
> 2025-05-07 haskell-charsetdetect-ae 1.1.0.4-7.1
> 2025-05-09 haskell-ogma-cli 1.7.0-2
> 2025-05-24 ijs 0.35-15.3
>
> ( If somebody could double-check the above, it would be great )
double-checked (haskell-ogma-{core,cli} are non-free)
> Now we could try building with old cdbs and new cdbs to see
> if there is any difference using diffoscope.
>
> Most packages in the above list are arch:all so there is very little chance
> to be misbuilt due to config.guess/config.sub.
>
> Maybe the other block that was also removed:
>
> config_rpath := $(shell find $(DEB_SRCDIR) \( -type f -or -type l \) -name
> config.rpath
>
> has more chances to create misbuilds.
if this is about rpath, isn't it going to be specific to arch:any builds
as well?
Breakdown per architecture:
udd=> select source, architecture from sources_uniq where release='sid'
and source in ('libiptcdata', 'vim-scripts', 'oss4', 'x2gobroker',
'dhelp', 'libmessage-passing-filter-regexp-perl',
'libmodule-install-copyright-perl',
'libmodule-install-doapchangesets-perl', 'libmodule-install-doap-perl',
'libmodule-install-rdf-perl', 'libmoosex-arrayref-perl',
'libpath-router-perl', 'librdf-trinex-serializer-mockturtlesoup-perl',
'librdf-vcard-perl', 'haskell-ogma-core', 'haskell-charsetdetect-ae',
'haskell-ogma-cli', 'ijs') order by architecture;
source | architecture
----------------------------------------------+--------------
dhelp | all
libmodule-install-doap-perl | all
libmodule-install-rdf-perl | all
libmoosex-arrayref-perl | all
libpath-router-perl | all
librdf-trinex-serializer-mockturtlesoup-perl | all
librdf-vcard-perl | all
vim-scripts | all
libmessage-passing-filter-regexp-perl | all
libmodule-install-copyright-perl | all
libmodule-install-doapchangesets-perl | all
haskell-ogma-cli | any
oss4 | any
haskell-charsetdetect-ae | any all
libiptcdata | any all
x2gobroker | any all
ijs | any all
haskell-ogma-core | any all
(18 rows)
> (my personal bet is that the impact is going to be zero).
I agree.
What is a bit painful here is that we would need to check binaries on
all architectures. Maybe it's easier to just bin-NMU them all...
Lucas
