Hi Bastien,
On 02/06/2024 14:38, Bastien Roucariès wrote:
Le dimanche 2 juin 2024, 11:17:33 UTC Sebastian Ramacher a écrit :
On 2024-02-02 17:21:43 +0000, Bastien Roucariès wrote:
Le vendredi 2 février 2024, 16:53:10 UTC Sebastian Ramacher a écrit :
Control: tags -1 moreinfo
Hi Bastien
On 2024-01-05 22:35:44 +0000, Bastien Roucariès wrote:
Package: release.debian.org
Severity: important
User: release.debian....@packages.debian.org
Usertags: transition
X-Debbugs-CC: ftpmas...@debian.org
Imagemagick will need a new major bump
I achieved to get imagemagick 7 build for experimental (it is only on salsa not
uploaded yet).
Every package include a version in the package name (except legacy package name
and perl*) so I plan to do some step by step migration, because it is mainly
coinstallable with imagemagick 6.
Why does this migration require co-instabillity with the old version?
This makes the transition overly complicated. Do you expect major
changes required in reverse dependencies of imagemagick's shared
library?
The problem is not the library but the command line interface that may need
change.
Librarry will break (I think here about php module that will need a update),
but it is treatable.
convert6 is not fully compatible with convert7
convert6 will be co installable with convert7 in order to test, and convert
will be provided by alternative system.
If they are not fully compatible, then alternatives are not an option.
They are 95% compatible
How many packages are we talking about? Have bugs been filed for
packages thar are not compatible with convert7?
The problem is chicken and eggs problem. If you could not test then you could
not report bug.
A least both should be in experimental for running a full archive rebuild
Not really. I also don't think the 3-step migration plan in experimental you're
proposing makes much sense.
What we need here is to know how many packages fail to build against imagemagick
7. And for that, the package doesn't even need to be in experimental. Also, you
don't need to perform a full archive rebuild, just rebuilding those packages
that build-depend on imagemagick would be enough.
Once that is done and we're ready to go, after going through experimental as
src:imagemagick (no renamed package), then we would start the transition and
scheduled the necessary binNMUs, and raise any bugs to serious.
That'd be the ideal scenario. But we don't know how feasible that is until we
know how many packages will need source changes, and for that we need the
results of the test rebuilds.
Cheers,
Emilio
