(Adrian: Added you to CCs per suggestion of Paul.)

Hi Paul,

On 2 January 2024 at 21:00, Paul Gevers wrote:
| Hi Dirk,
|
| On 02-01-2024 20:42, Dirk Eddelbuettel wrote:
| > | The Release Team considers packages that are out-of-sync between testing | > | and unstable for more than 30 days as having a Release Critical bug in
| >
| > I noticed that too over the last few weeks as I tend to keep an eye on my
| > aggregation at https://qa.debian.org/developer.php?login=edd@debian.org
|
| Nice. I wish every DD did that.
|
| > | This bug will trigger auto-removal when appropriate. As with all new
| > | bugs, there will be at least 30 days before the package is auto-removed. | >
| > Sure. Though that might be harsh / might affect other packages.
|
| They will be notified of the autoremoval automatically and can help you
| fix the situation. If there's work in progress, you can delay the
| autoremoval by pinging this bug, that resets the timer.
|
| > We may want to consider exempting i386 as a build arch if that is possible. |
| Well, if you really can't support i386 anymore (we expect from DD to
| support as many architectures as is *reasonably* possible), you should
| arrange for the removal of the i386 package, including all reverse i386
| dependencies. It would be good to coordinate this with your reverse
| dependencies (at least inform them). In the end removal happens by
| filing appropriate RM bugs against the ftp.debian.org pseudo package.

Ok. I can do that. I just look at 'rdepends' for r-cran-rjava and it is only five packages. That seems fair.

| > | If you believe your package is unable to migrate to testing due to
| > | issues beyond your control, don't hesitate to contact the Release Team.
| >
| > :wave:
|
| FTBFS of your own package is what I consider to be in your control (this
| text is part of the template I use). Either you fix the issue, or you
| decide to no long support i386 with your package, but you'll need to
| coordinate with your reverse dependencies. The removal happens by
| ftp-master once you file the appropriate bugs.
|
| > This is an R package, and R no longer releases on i386 meaning upstream may | > not have checked / may not be receptive. See eg [1] for the CRAN state of the
| > package. No i386 there.
| >
| > I am not sure what else to do besides simply saying 'no longer builds on i386'.
|
| Maybe contact i386 porters for help creating a patch? (We have one:
| Adrian Bunk).

Good idea. Have CC'ed Adrian to see if he wants to jump in.

| Having said all that, most of our upstreams don't support (for some
| value of support) all the architectures that we support. Still we expect
| from DD to put in *reasonable* effort to support their packages on our
| architectures. So, the call to drop an architecture from the supported
| list is yours to make as a maintainer.

It is not easy to strike the right balance, ie for m68k we 'hang on' for a
long time as we had motivated maintainers / porters / developers. Not sure we had users :)

For i386 we have been patient too. The hardware has been EOL for some time
and most projects have ceased explicit support. That is a fair sign.

If someone wants to help, I am happy to play along. But if not, I think for a 'somewhat marginal leaf-alike' dependency such as rJava aka r-cran-rjava removing i386 support is defensible. We only support approx 1k out 20k CRAN packages so users are accustomed to having to go elsewhere anyway. I packaged rJava nearly 20 years ago because it is a 'difficult' package for many users and our integration helps. I still maintain it for the same reason, even if Java is also way more marginal within R now. So for i386 the end may be coming.

Cheers, Dirk


| Paul
| x[DELETED ATTACHMENT OpenPGP_signature.asc, application/pgp-signature]

--
dirk.eddelbuettel.com | @eddelbuettel | edd@debian.org

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)