Currently manuscripts developed in inst/doc/pubs or something like that; rather obtuse. Should manuscripts just be treated like other vignettes?
Default behavior for knitr .Rmd vignette is to build HTML. Could add a Makefile (called after knitr compiles the vignette) to recompile into md and then into tex/pdf using pandoc, but would then assume/require pandoc to be installed to build the package. Desirable to separate dependencies needed for automatic and remote build workflow from dependencies a user actually needs.
Looking at knitr::pandoc, e.g. knitr pandoc demo. Why the pandoc wrapper in the first place? Why not build on pander?
Looking at pander. Presumably user still needs to install pandoc. (Yup, INSTALL in package root and linked to CRAN points them to Pandoc's binaries).
Would it help to declare ruby in a list of languages in my notebook .travis.yml? And/or particular versions, e.g.
cboettig
commented
10 years ago
Currently manuscripts developed in
inst/doc/pubsor something like that; rather obtuse. Should manuscripts just be treated like other vignettes?Default behavior for knitr
.Rmdvignette is to build HTML. Could add a Makefile (called after knitr compiles the vignette) to recompile into md and then into tex/pdf using pandoc, but would then assume/require pandoc to be installed to build the package. Desirable to separate dependencies needed for automatic and remote build workflow from dependencies a user actually needs.knitr::pandoc, e.g. knitr pandoc demo. Why the pandoc wrapper in the first place? Why not build on pander?pander. Presumably user still needs to install pandoc. (Yup, INSTALL in package root and linked to CRAN points them to Pandoc's binaries).Would it help to declare ruby in a list of languages in my notebook .travis.yml? And/or particular versions, e.g.