1.13.0 release checks

[mattdowle]

• Rerun all revdeps:
  already in error on CRAN with 1.12.8: inlmisc mlr3proba Require rgbif strataG wyz.code.metaTesting
  can't get metaMix past "Invalid MIT-MAGIC-COOKIE-1 key" locally
  statcanR fails locally with 1.12.8 on "error:1414D172:SSL routines:tls12_check_peer_sigalg:wrong signature type"
  skynet fails locally with download problem Download error from R CMD check ropensci/skynet#6
  traitdataform fails locally with download problem, reruns same issue
  the following revdep errors are caused by data.table upgrade (just 9 packages out of 867 👍) :
  (tick when either fixed in data.table, or we've heard from the maintainer ok to proceed)
• eplusr eplusr error on upgrade data.table #4612
• expss set2key error with next release of data.table 1.13.0 gdemin/expss#67
• nc nc error on upgrade data.table #4614 Tests broken by upcoming data.table release tdhock/nc#15 and emailed
• paradox paradox error on upgrade data.table #4607
• popEpi popEpi error on upgrade data.table #4615
• socialmixr socialmixr error on upgrade data.table #4616
• tableMatrix tableMatrix error on upgrade data.table #4617
• maditr set2key error with next release of data.table 1.13.0 gdemin/maditr#14
• memochange memochange warning on upgrade data.table #4618 fix vignette issue broken by fread upstream KaiWenger/memochange#13
• CRAN_Release.cmd (e.g. UBSAN, ASAN, rchk rchk results to work through #4625, gctorture strict gctorture results to work through #4628)
• pass win-builder
• CRAN 'additional issues' pass clang-ubsan #4624
• News item review and tidy
• Codefreeze
• @renkun-ken go-first run please (no error with aa60871)
• Repeat win-builder 3*OK
• Start final rerun latest revdeps refreshed. CRAN runs revdep impact analysis as well before accepting, and they often find things I don't locally. Hence don't wait for this to complete.
• Submit to CRAN (several opportunities to halt after this, so please shout if something comes up)
• CRAN pre-test success email received and ok
• Bump dev version
• Copy and paste revdep.R::status() output as new comment to this issue
• CRAN revdep status impact results email received
• Matt replied to confirm and explain differences to submission note
• CRAN email received to confirm accepted
• Add publish date to v1.13.0 title in NEWS.md
• Update version and date in Installation wiki heading
• Tweet to #rdatatable https://twitter.com/MattDowle/status/1286736488835801089
• actel and spatsoc follow up TZ=UTC causes actel tests to fail with data.table 1.13.0 hugomflavio/actel#46 TZ=UTC causes spatsoc tests to fail with data.table 1.13.0 ropensci/spatsoc#32

Based on 1.12.4 release check list: #3863

[mattdowle]

Revdep status on submit. Reasons in the top comment of this issue.

> status()
Installed data.table to be tested against: 1.13.0 2020-07-22 23:00:19 
CRAN:
 ERROR   :  12 : actel expss mlr3proba nc Require rgbif skynet spatsoc statcanR strataG traitdataform
                 wyz.code.metaTesting 
 WARNING :   2 : memochange metaMix 
 NOTE    : 292 
 OK      : 564 
 TOTAL   : 870 / 870 
 
BIOC:
 ERROR   :  19 : adductomicsR CATALYST cellbaseR cTRAP DMCFB DominoEffect ImmuneSpaceR LOLA LowMACA maser
                 methylPipe MouseFM PGA psichomics QuartPAC R3CPET rfaRm RGMQL TFutils 
 WARNING :  40 : artMS BEclear biscuiteer BrainSABER bsseq CellNOptR cicero cmapR CONFESS contiBAIT
                 crossmeta CytoML DiscoRhythm ELMER eQTL fcoex flowWorkspace geneplast geneXtendeR GenoGAM
                 genomation GenomicScores GmicR HMMcopy iCNV methrix MinimumDistance MMAPPR2 MSnID Onassis
                 phantasus pwOmics RiboProfiling S4Vectors sangeranalyseR SEtools signatureSearch SPONGE
                 TitanCNA ViSEAGO 
 NOTE    : 108 
 OK      :  40 
 TOTAL   : 207 / 207 
 
TOTAL          : 1077

actel and spatsoc were new in the final rerun I started at the same time as submitting to CRAN. They were not raised by CRAN's revdep checks so I did not halt CRAN release. It turned out they failed due to me running revdep checks using TZ=UTC which has been in revdep.R for a few years now. This means CRAN check machines run in their local timezone. Links to issues for actel and spatsoc in the top comment. What happened most likely, thinking about it now, is that when revdep checks complete, I often rerun the failed ones manually in case they failed just because of resource issues, running multiple package checks in parallel, or internet download issues. That rerun one-by-one often clears a few. When I did that rerun I probably just typed Rdevel CMD check <tar.gz> without the TZ=UTC that the revdep run() does. Hence I missed actel and spatsoc. Fortunately, they are not impacted on CRAN and are still passing. Interestingly, what this means is that if we were to change the default of tz= to UTC, only 2 packages on CRAN would be impacted: actel and spatsoc. Because it turns out I have just tested that since the revdep checks ran all packages with TZ=UTC.

I did not look at Bioconductor this time. There were repeated requests from Kurt and Uwe to update on CRAN as soon as possible because data.table's error status (caused by breaking change in R) was causing downstream issues. Bioconductor still has too many errors and warnings that, collectively and in general relative to CRAN, Bioconductor maintainers have not fixed for many years now despite me communicating many times with them. This makes it too time consuming to filter through what is caused by data.table and what is not. Responses usually suggest ways that I can do that (e.g. rerunning with 1.12.8 and only looking at the deltas) ... which is more work. This seems unreasonable after so many years now of Bioconductor maintainers in general not fixing errors and in particular warnings.

fail.log