****************************
release checklist for CHNOSZ
****************************

- uncomment CSL line in Rmd files (commented for building on R-Forge)
  csl: elementa.csl

- compile vignettes with release version number and date

- run examples() and demos() to check their output

- check output of demo("sources") to make sure all data sources are cited

- recreate extdata/thermo/obigt_check.csv after all data updates:
  co <- check.obigt()
  write.csv(co, "obigt_check.csv", row.names=FALSE, na="")

- check that uniprot.aa() works with current UniProt web pages

- run R_PAPERSIZE=letter R CMD Rd2pdf chnosz/
  and fix any lines truncated by page margins

- run R CMD check using R compiled without long doubles (emulating Solaris checks on CRAN)
  (CFLAGS=-ffloat-store ./configure --disable-long-double)

- build the package on the source directory;
  try both qpdf and ghostscript to compact vignettes:
  R CMD build --compact-vignettes=both chnosz/

OTHER THINGS TO CHECK:

- remove all "<<-" from sources (leftover from debugging)

- update list of documentation topics in examples() with any new ones

- especially on Windows: anim.*() produce pngs / movies (ImageMagick dependency)?

- check for stale URLs in Rd files

- clean up all TODO, XXX, etc. in Rd files (and hopefully R files too)

- update .Rinstignore with any new/deleted files

- ensure all Rd files have \keyword{...} or \concept{...}
  - keywords used are standard R keywords: package, utilities
  - concepts are as follows:
    - Thermodynamic data (thermo, extdata)
    - Primary thermodynamic modeling (commonly used functions e.g. subcrt)
    - Secondary thermodynamic modeling (less often-used functions e.g. revisit)
    - Extra thermodynamic modeling (out-of-workflow functions e.g. transfer)
    - Protein thermodynamic modeling (protein-specific functions e.g. ionize.aa)

- although Rd files could use the \doi macro, it generates a largish (ca. 265K)
  build/partial.rdb file; just use \url for now.
