Re: R-beta: auto save

Martin Maechler (
Fri, 30 May 97 09:00:58 +0200

Date: Fri, 30 May 97 09:00:58 +0200
Message-Id: <9705300700.AA03702@>
From: Martin Maechler <>
In-Reply-To: <> (message from
Subject: Re: R-beta: auto save

>>>>> "Rob" == Robert Gentleman <> writes:

    Rob> I'm just about to put an autosave feature into R as per requests...


    Rob> My current thinking is that it should be an option
    Rob> so that it is setable (in the profile if you want).

sounds very good.

    Rob> One simply specifies the number of commands that are executed
    Rob> between calls to save.

Let's call this number  ASI ( = Auto_Save_Interval)
The question is  how the default value for ASI is to be determined.

I think that ASI = Inf  for  'BATCH' mode  and when using source(.),
but  ASI < Inf  even if the user does not have a ~/.Rprofile.

    Rob> It won't be incremental; I'll just dump the output
    Rob> of ls().  I think that the file name should not be .RData but
    Rob> rather something like .RAutoSave (so it's there if you need it but
    Rob> you don't get it on startup unless you ask for it).

The following is a different proposal,
maybe more intuitive for "novices" :

A regular   q(ask)   probably should remove  .RAutoSave:
			[ if ask='yes', we have a new .RData anyway,
			  and otherwise, we wanted to rm(.) things ]
Then, upon startup, when there's an  .RAutoSave newer than .RData,
it must have been from a "crash" (non-regular termination of R),
and the user would be asked if (s)he wanted to use  .RAutoSave instead of

r-help mailing list -- Read
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: