[R-pkg-team] Bug#1002525: r-cran-tmb: autopkgtest needs update for new version of rmatrix: Package version inconsistency detected.

[Dirk Eddelbuettel]

Hi Nilesh,

On 26 December 2021 at 00:20, Nilesh Patra wrote:
| control: reassign -1 r-cran-tmb/1.7.22-1
| 
| Hi Dirk,
| 
| On Thu, 23 Dec 2021 14:30:10 -0600 Dirk Eddelbuettel <edd at debian.org> wrote:
| >                         pass            fail
| > | rmatrix                from testing    1.4-0-1
| > | r-cran-tmb             from testing    1.7.22-1
| > | all others             from testing    from testing
| > | 
| > | I copied some of the output at the bottom of this report. It seems the 
| > | binary embeds the version of rmatrix it's build against without 
| > | declaring proper versioned Dependencies to reflect that.
| > 
| > Yes. Apparently a design decision of the (R package) TMB package imposed
| > after the slight accidents that append with the (R package) Matrix (aka
| > rmatrix for us) during its last ABI/API change (at version 1.3.0 and after).
| > Not a lot we can do, other than to patch it out in TMB.
| 
| Interesting. OK, I did the needful and uploaded. rmatrix should migrate soonish.
| 
| > Nothing I can do (as maintainer of r-cran-matrix built of source package
| > rmatrix aka CRAN package Matrix.
| 
| What you can do is inform people about it, by sending an email to the mailing list.
| Someone would take that up soonish.
| As per your words, and as per the autopkgtest problem, it needs a re-compilation, with
| every new rmatrix, and that does not happen by default with buildd machines, right.

It's a fairly rare case. I had it myself with Rcpp (version 1.0.7) too.

What we might need _within the R ecosystem_ is the ability to push binary
rebuilds through. 

| > | Currently this regression is blocking the migration of rmatrix to 
| > | testing [1]. Of course, rmatrix shouldn't just break your autopkgtest 
| > | (or even worse, your package), but it seems to me that the change in 
| > | rmatrix was intended and your package needs to update to the new situation.
| > 
| > None of that, AFAIK, comes from R package Matrix. It is just TMB.
| 
| Yep.
| 
| > Coupled with what happens around here with our ability to not keep packages
| > aligned with their CRAN versions.
| 
| For this particular bug, tmb is already at the latest version, so it has got nothing to
| do with this bug report.

Yes. I guess it is the forced rebuild we need.

I wonder if the TMB maintainer should have resorted to  Matrix (>= 1.4.0)  ?

| If I talk about it otherwise then, I am usually (personally) trying to keep everything to CRAN-latest
| and Andreas helps with that. There are some blockers in the process for instance introduction of new cran
| dependencies that are un-packaged. So we quickly package that and upload to NEW, but then,
| there is nothing we can do in those cases except waiting action for accept from the FTP-team.
| Despite this, the number of cran-'non'-latest had been well below 20 (IIRC it was max 10) for past
| several months.
| 
| It has now unfortunately increased a little (but manageable) because we have less time.
| At the end of the day, we are all 'volunteers'
| 
| > | If this is a real problem in your package (and not only in your 
| > | autopkgtest), the right binary package(s) from rmatrix should really add 
| > | a versioned Breaks on the unfixed version of (one of your) package(s). 
| > 
| > I don't think I agree. Matrix does nothing here. You appear to be shooting a
| > messenger.
| 
| Actually, that's sort of a templated reply that you see above which is reasonable since there are so many similar bugs,
| if you see other bugs that break autopkgtests, you will find the wording to be awfully similar. For example:
| 
| https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=996279
| 
| Hence, we need to re-assign and check accordingly.

As I mentioned before, I cannot not help but think these autopkgtests _for
CRAN packages_ add nothing (as CRAN does for all tests for us already), but
create extra work (when we slip away from CRAN pairings and create our own
bugs).

But we're all adults and as you all feel it adds value, you keep doing
it. Such is life. We're volunteers and if that is what you all do then I am
still thankful for all the other work ;-)

| Lastly, Merry Christmas! :)

Thanks!  Same to you.

Dirk
 
| Kind Regards,
| Nilesh
| x[DELETED ATTACHMENT signature.asc, application/pgp-signature]

-- 
https://dirk.eddelbuettel.com | @eddelbuettel | edd at debian.org