On 04/07/2019 12:32 p.m., Spencer Brackett wrote:> Hello again, > > I might be repeating myself here, so my apologies, but do I have to run a > script file from my R Studio to reimplement my previous work for a given > project.... so to start up where I left off.... or is opening up R and, > with my global environment automatically reloading as it was when I last > worked on, sufficient?Saving your workspace when you quit is a common default, but it is generally a bad idea. Old junk collects in there, and makes new results harder to debug. A better workflow is to never save the whole workspace. If you have just computed some object(s) and the computation took so long you don't want to repeat it, then save just a minimum, and load them later in a new session. A particularly dangerous situation happens if you sometimes save your workspace and sometimes don't. You can end up with situations like this: Session 1: compute some random values. Save the workspace, including the random number key. Session 2: automatically load the saved workspace. Compute some new random values. Quit without saving the workspace. Session 3: automatically load the saved workspace from Session 1, including the random number seed. Any random values computed in this session could be identical to the values in Session 2, because they are starting with the same seed. If you don't have a saved workspace to load, you end up with a blank slate, and the random number key is generated based on time of day and process number, so is almost certainly different in every session. (Sometimes you want a repeated seed for reproducibility, but it's always bad when you're surprised by one.) Duncan Murdoch
Thank you for the clarification. So should I not rely on importing a saved environment from now on? I am currently experiencing some difficulties with reproducing the output (aka the objects listed in my environment), which is why I was trying to load them all at once. Best, Spencer On Thu, Jul 4, 2019 at 1:24 PM Duncan Murdoch <murdoch.duncan at gmail.com> wrote:> On 04/07/2019 12:32 p.m., Spencer Brackett wrote: > > Hello again, > > > > I might be repeating myself here, so my apologies, but do I have to > run a > > script file from my R Studio to reimplement my previous work for a given > > project.... so to start up where I left off.... or is opening up R and, > > with my global environment automatically reloading as it was when I last > > worked on, sufficient? > > > Saving your workspace when you quit is a common default, but it is > generally a bad idea. Old junk collects in there, and makes new results > harder to debug. > > A better workflow is to never save the whole workspace. If you have > just computed some object(s) and the computation took so long you don't > want to repeat it, then save just a minimum, and load them later in a > new session. > > A particularly dangerous situation happens if you sometimes save your > workspace and sometimes don't. You can end up with situations like this: > > Session 1: compute some random values. Save the workspace, including > the random number key. > > Session 2: automatically load the saved workspace. Compute some new > random values. Quit without saving the workspace. > > Session 3: automatically load the saved workspace from Session 1, > including the random number seed. Any random values computed in this > session could be identical to the values in Session 2, because they are > starting with the same seed. > > If you don't have a saved workspace to load, you end up with a blank > slate, and the random number key is generated based on time of day and > process number, so is almost certainly different in every session. > (Sometimes you want a repeated seed for reproducibility, but it's always > bad when you're surprised by one.) > > Duncan Murdoch >[[alternative HTML version deleted]]
If you can't reproduce them, how do you know they are correct? On July 4, 2019 11:34:53 AM PDT, Spencer Brackett <spbrackett20 at saintjosephhs.com> wrote:>Thank you for the clarification. So should I not rely on importing a >saved >environment from now on? I am currently experiencing some difficulties >with >reproducing the output (aka the objects listed in my environment), >which is >why I was trying to load them all at once. > >Best, > >Spencer > >On Thu, Jul 4, 2019 at 1:24 PM Duncan Murdoch ><murdoch.duncan at gmail.com> >wrote: > >> On 04/07/2019 12:32 p.m., Spencer Brackett wrote: >> > Hello again, >> > >> > I might be repeating myself here, so my apologies, but do I have >to >> run a >> > script file from my R Studio to reimplement my previous work for a >given >> > project.... so to start up where I left off.... or is opening up R >and, >> > with my global environment automatically reloading as it was when I >last >> > worked on, sufficient? >> >> >> Saving your workspace when you quit is a common default, but it is >> generally a bad idea. Old junk collects in there, and makes new >results >> harder to debug. >> >> A better workflow is to never save the whole workspace. If you have >> just computed some object(s) and the computation took so long you >don't >> want to repeat it, then save just a minimum, and load them later in a >> new session. >> >> A particularly dangerous situation happens if you sometimes save your >> workspace and sometimes don't. You can end up with situations like >this: >> >> Session 1: compute some random values. Save the workspace, >including >> the random number key. >> >> Session 2: automatically load the saved workspace. Compute some new >> random values. Quit without saving the workspace. >> >> Session 3: automatically load the saved workspace from Session 1, >> including the random number seed. Any random values computed in this >> session could be identical to the values in Session 2, because they >are >> starting with the same seed. >> >> If you don't have a saved workspace to load, you end up with a blank >> slate, and the random number key is generated based on time of day >and >> process number, so is almost certainly different in every session. >> (Sometimes you want a repeated seed for reproducibility, but it's >always >> bad when you're surprised by one.) >> >> Duncan Murdoch >> > > [[alternative HTML version deleted]] > >______________________________________________ >R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see >https://stat.ethz.ch/mailman/listinfo/r-help >PLEASE do read the posting guide >http://www.R-project.org/posting-guide.html >and provide commented, minimal, self-contained, reproducible code.-- Sent from my phone. Please excuse my brevity.