Active bindings in attached environments

Hi,

Is this expected behavior for active bindings in attached
environments, or is this a bug:
e <- new.env()
makeActiveBinding('x',function() 'foo',e)
ls(e)
[1] "x"
attach(e)
search()
[1] ".GlobalEnv"        "e"                 "package:graphics"
[4] "package:grDevices" "package:datasets"  "package:utils"
[7] "package:methods"   "Autoloads"         "package:base"
x
function() 'foo'

Should this print 'foo' ? The following works as I would expect:
with(e,x)
[1] "foo"

but this doesn't:
f <- function() x
f()
function() 'foo'

However, changing the environment of f does:
environment(f) <- e
f()
[1] "foo"

Jeff
http://biostat.mc.vanderbilt.edu/JeffreyHorner
I think it is related to the call to duplicate in do_attach:

defineVar(TAG(p), duplicate(CAR(p)), s);

You can circumvent it by using "importIntoEnv", as in:

 > e
<environment: 0x9070064>
 > e <- new.env()
 > makeActiveBinding('x',function() 'foo',e)
 > f <- new.env()
 > attach( f, pos = 2 )
 > .Internal(importIntoEnv( as.environment(2), ls(e), e, ls(e) ))
NULL
 > x
[1] "foo"

Romain
Hi,

Is this expected behavior for active bindings in attached
environments, or is this a bug:

e<- new.env()
makeActiveBinding('x',function() 'foo',e)
ls(e)
[1] "x"
attach(e)
search()
[1] ".GlobalEnv"        "e"                 "package:graphics"
[4] "package:grDevices" "package:datasets"  "package:utils"
[7] "package:methods"   "Autoloads"         "package:base"
x
function() 'foo'

Should this print 'foo' ? The following works as I would expect:

with(e,x)
[1] "foo"

but this doesn't:

f<- function() x
f()
function() 'foo'

However, changing the environment of f does:

environment(f)<- e
f()
[1] "foo"

Jeff

Romain Francois
Professional R Enthusiast
+33(0) 6 28 91 30 30
http://romainfrancois.blog.free.fr
|- http://tr.im/BcPw : celebrating R commit #50000
|- http://tr.im/ztCu : RGG #158:161: examples of package IDPmisc
`- http://tr.im/yw8E : New R package : sos
Also, active bindings on the search path seem to do only half of the job:

 > e <- new.env()
 > attach( e )
 > makeActiveBinding( "x", function(val) if(missing(val)) "get" else 
"set", as.environment(2) )
 > x
[1] "get"
 > x <- 3
 > x
[1] 3

Romain

I think it is related to the call to duplicate in do_attach:

defineVar(TAG(p), duplicate(CAR(p)), s);

You can circumvent it by using "importIntoEnv", as in:

 > e
<environment: 0x9070064>
 > e <- new.env()
 > makeActiveBinding('x',function() 'foo',e)
 > f <- new.env()
 > attach( f, pos = 2 )
 > .Internal(importIntoEnv( as.environment(2), ls(e), e, ls(e) ))
NULL
 > x
[1] "foo"

Romain

On 11/05/2009 04:53 PM, Jeffrey Horner wrote:
Hi,

Is this expected behavior for active bindings in attached
environments, or is this a bug:

e<- new.env()
makeActiveBinding('x',function() 'foo',e)
ls(e)
[1] "x"
attach(e)
search()
[1] ".GlobalEnv" "e" "package:graphics"
[4] "package:grDevices" "package:datasets" "package:utils"
[7] "package:methods" "Autoloads" "package:base"
x
function() 'foo'

Should this print 'foo' ? The following works as I would expect:

with(e,x)
[1] "foo"

but this doesn't:

f<- function() x
f()
function() 'foo'

However, changing the environment of f does:

environment(f)<- e
f()
[1] "foo"

Jeff

Romain Francois
Professional R Enthusiast
+33(0) 6 28 91 30 30
http://romainfrancois.blog.free.fr
|- http://tr.im/BcPw : celebrating R commit #50000
|- http://tr.im/ztCu : RGG #158:161: examples of package IDPmisc
`- http://tr.im/yw8E : New R package : sos
An embedded and charset-unspecified text was scrubbed...
Name: not available
URL: <https://stat.ethz.ch/pipermail/r-devel/attachments/20091105/bcfe98aa/attachment.pl>
Hi,

Is this expected behavior for active bindings in attached
environments, or is this a bug:

e <- new.env()
makeActiveBinding('x',function() 'foo',e)
ls(e)
[1] "x"
attach(e)
search()
[1] ".GlobalEnv" ? ? ? ?"e" ? ? ? ? ? ? ? ? "package:graphics"
[4] "package:grDevices" "package:datasets" ?"package:utils"
[7] "package:methods" ? "Autoloads" ? ? ? ? "package:base"
x
function() 'foo'

Should this print 'foo' ? The following works as I would expect:

with(e,x)
[1] "foo"

but this doesn't:

f <- function() x
f()
function() 'foo'

However, changing the environment of f does:

environment(f) <- e
f()
[1] "foo"
Actually, it is my understanding of attach() which is the bug. The
attach documentation clearly states that a copy of the object is
attached, not the object itself.

Thanks,

Jeff
--
http://biostat.mc.vanderbilt.edu/JeffreyHorner
An explanation of this would be nice in ?makeActiveBinding, e.g.,

NOTE:

When an environment is attach()'ed to the search path, any active 
bindings in it are not preserved as active bindings.  This happens 
because attach() actually adds a new environment to the search path, and 
copies objects into it.  Thus, active bindings are evaluated and their 
values copied into the attached environment.

(Is this explanation correct?)

-- Tony Plate

(PS: Thunderbird wants to spell-correct "makeActiveBinding" to 
"vindictivenesses", which seems an amusingly appropriate allusion to the 
nature of this "gotcha" :-)
On Thu, Nov 5, 2009 at 9:53 AM, Jeffrey Horner <jeffrey.horner at gmail.com> wrote:

Hi,

Is this expected behavior for active bindings in attached
environments, or is this a bug:

e <- new.env()
makeActiveBinding('x',function() 'foo',e)
ls(e)

[1] "x"

attach(e)
search()

[1] ".GlobalEnv"        "e"                 "package:graphics"
[4] "package:grDevices" "package:datasets"  "package:utils"
[7] "package:methods"   "Autoloads"         "package:base"

x

function() 'foo'

Should this print 'foo' ? The following works as I would expect:

with(e,x)

[1] "foo"

but this doesn't:

f <- function() x
f()

function() 'foo'

However, changing the environment of f does:

environment(f) <- e
f()

[1] "foo"

Actually, it is my understanding of attach() which is the bug. The
attach documentation clearly states that a copy of the object is
attached, not the object itself.

Thanks,

Jeff
--
http://biostat.mc.vanderbilt.edu/JeffreyHorner

______________________________________________
R-devel at r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel