Dear developeRs, my package FrF2.catlg128 holds large catalogues and is supposed to gain additional ones. All the catalogues are intended for the user. So far, the catalogues were stored in the data directory, and LazyData was "no". I understand that this is not considered wise any more (if it ever was), so that I want to change to LazyData "yes" with the next release (which will also get some additional catalogues). I have tried out using separate data files in the data directory (like before) and one sysdata.rda file in the R directory (exporting all catalogues from the namespace); there is a large difference in the installed sizes between those two ways: the approach with sysdata.rda uses only about half the size of the separate data files approach (5.6 Mb vs 11.7Mb). As I would like to be able to query the available data in the package via data(package="FrF2.catlg128")) even before the package is loaded, I want to have a data directory with a datalist file in there. This appears to be compatible with using a sysdata.rda file in the R directory. (From a tidyness point of view, I would prefer the data file to sit in the data directory as well; however, that about doubles the installed size again (11.4 vs 5.6Mb) even if I use just the one sysdata.rda file.) Regarding the installed package size, the best option is obviously one sysdata.rda file in the R directory, but I want the datalist file for the reason given above. A data directory without data files throws a warning, so that I have to include a dummy data file (and documentation for it) for allowing me to have a datalist file. Finally my questions: Is there a better way to achieve what I am looking for? And if not: is there any reason against combining a sysdata.rda file in the R directory with a datalist file (that lists the data from the sysdata.rda file) in the data directory, be it policy-wise or perhaps in terms of memory usage within an R session? Best regards, Ulrike
sysdata.rda vs. rda files in data directory
2 messages · Ulrike Grömping
4 days later
Dear all, for everybody's benefit, here is a brief summary of what I learnt from Uwe Ligges. There is a convincing reason against combining a data directory with a datalist file only (the way for retrieving a list of the data available without the package loaded) with a sysdata.rda file in the R directory (the only way for achieving a small installed size): the command data(filename) will not work on the names listed in the datalist file (I thought otherwise), which would be sort of odd and unexpected. Therefore, I will go for the small installed size (=sysdata.rda file) without a data directory. Best, Ulrike -- View this message in context: http://r.789695.n4.nabble.com/sysdata-rda-vs-rda-files-in-data-directory-tp4662302p4662661.html Sent from the R devel mailing list archive at Nabble.com.