search

BAAQMD / fmatch

Fast string matching with limited wildcard support
0 stars 0 forks source link

R CMD CHECK fails when building vignette #5

Open arilamstein opened 3 years ago

arilamstein commented 3 years ago

There are no errors when checking the package proper. But when it gets to the Vignette it generates this error:

Error: package or namespace load failed for 'tbltools' in loadNamespace(j <- i[[1L]], c(lib.loc, .libPaths()), versionCheck = vI[[j]]):
   there is no package called 'data.table'
  Quitting from lines 13-18 (fmatch_performance.Rmd) 
  Error: processing vignette 'fmatch_performance.Rmd' failed with diagnostics:
  package 'tbltools' could not be loaded
  --- failed re-building ‘fmatch_performance.Rmd’

  SUMMARY: processing the following file failed:
    ‘fmatch_performance.Rmd’

  Error: Vignette re-building failed.
  Execution halted

Note that tbltools is on the server. It is not explictly mentioned in Suggests:, but I believe it is imported by other packages which the package mentions in the DESCRIPTION file. So it is not exactly clear why the package cannot be found.

One possible theory is that R CMD CHECK might be doing some kind of strict verification to make sure that each package can be installed. Note, however, that the tbltools repository is private on github: https://github.com/BAAQMD/tbltools. So tbltools might not be able to be "found" under that criteria.

Another theory is that fmatch and tbltools both import/suggest/depend on each other, which might be causing a problem.

dholstius commented 3 years ago

As we discussed today (2021-04-06), I'm OK with closing this, given that:

... so long as we capture the error in a separate issue. (Once that happens, please link the new issue here, and then close this one—thanks!)

arilamstein commented 3 years ago

As we discussed today (2021-04-06), I'm OK with closing this, given that:

  • We both get the same error when running check(); and
  • For our present purposes, it seems harmless

... so long as we capture the error in a separate issue. (Once that happens, please link the new issue here, and then close this one—thanks!)

Hi @dholstius. I'm slightly confused by your message. I created this issue just as a bookmark to acknowedge the issue. It sounds like you want me to close this (which is fine). What confuses me is that it sounds like you also want me to open up a new issue that has the same information.

dholstius commented 3 years ago

Ah. I think the issue just needs to be reframed – maybe just retitled. To me it's not about the vignette; it's about this weirdness we have been running into, in more than one context, where R complains that a package cannot be loaded during check, when it can be loaded via library in an interactive session. I'd like to capture that, and figure out what the root cause is. I think it might not just be specific to data.table. And I don't feel like I understand it.