Write org variables into own R environment instead of .GlobalEnv

Write org variables into own R environment instead of .GlobalEnv

[Rainer M Krug]

[-- Attachment #1: Type: text/plain, Size: 1832 bytes --]

Hi

I just send a patch which changes the behaviour of how org variables are
treated in R. At the moment, org variables are simply stored in the
.GlobalEnv which means, that all show up in the variable listing (which
can get cluttered when having many variables), they can accidentaly be
deleted and not be restored from within R and saving all the variables
into from R to make them available after tangling is not that easy.

Therefore the patch writes the variables into their own environment
(which I simply called org) and locks the environment and the
bindings. This means, that the actual variable values are always in the
environment and can not be accidentally deleted. As the environment is
attached to the search path, they are accessible as before, but they do
not clutter the workspace and do not show up separately e=when using
ls() to list the contents of the workspace. When using ls(org) all can
be seen.

They can still be "overwritten", but but this only creates a new
variable of the ame name in the .GlobalEnv which is simply hidint the
original variable passed from org. The original variable can be accessed
via org$VARIABLENAME. When removing the defined variable VARIABLENAME
via rm(VARIABLENAME), the value passed from org is back.

In addition, all variables can be easily be saved to disk by using
save(org, file="/PATH/TO/FILE.Rdata") and reloaded with
(load("/PATH/TO/FILE.Rdata")). This mechanism could actually be included
into tangling as the default mechanism to load the variables to avoid
cluttering the code with all the assignment commands of the org
variables.

The patch is not yet extensively tested.
Please provide some feedback about the idea and implementation,

Cheers,

Rainer

-- 
Rainer M. Krug
email: RMKrug<at>gmail<dot>com
PGP: 0x0F52F982

[-- Attachment #2: Type: application/pgp-signature, Size: 494 bytes --]

Re: Write org variables into own R environment instead of .GlobalEnv

[Andreas Leha]

Hi Rainer,

Rainer M Krug <Rainer@krugs.de> writes:

> Hi
>
> I just send a patch which changes the behaviour of how org variables are
> treated in R. At the moment, org variables are simply stored in the
> .GlobalEnv which means, that all show up in the variable listing (which
> can get cluttered when having many variables), they can accidentaly be
> deleted and not be restored from within R and saving all the variables
> into from R to make them available after tangling is not that easy.
>
> Therefore the patch writes the variables into their own environment
> (which I simply called org) and locks the environment and the
> bindings. This means, that the actual variable values are always in the
> environment and can not be accidentally deleted. As the environment is
> attached to the search path, they are accessible as before, but they do
> not clutter the workspace and do not show up separately e=when using
> ls() to list the contents of the workspace. When using ls(org) all can
> be seen.
>
> They can still be "overwritten", but but this only creates a new
> variable of the ame name in the .GlobalEnv which is simply hidint the
> original variable passed from org. The original variable can be accessed
> via org$VARIABLENAME. When removing the defined variable VARIABLENAME
> via rm(VARIABLENAME), the value passed from org is back.
>
> In addition, all variables can be easily be saved to disk by using
> save(org, file="/PATH/TO/FILE.Rdata") and reloaded with
> (load("/PATH/TO/FILE.Rdata")). This mechanism could actually be included
> into tangling as the default mechanism to load the variables to avoid
> cluttering the code with all the assignment commands of the org
> variables.
>
> The patch is not yet extensively tested.
> Please provide some feedback about the idea and implementation,

Currently I lack the time to test the patch.  So, this is just a
(not-so-helpful) feedback on the idea:  Seems to be very good!

Regards,
Andreas