Hello out there,
sorry if this is something obvious, but I don't know if I can trust my
eyes. I have been sweating several hours over the "problem" to get a
command executed inside a namespace - only to find out, that the cause
seems to be something totally different.
Consider this screen dump (from within TKCon):
(NEW2) 136 % catch {namespace delete sim}

0
(NEW2) 137 % namespace eval sim {}
(NEW2) 138 % namespace eval sim {set str "A string"}
A string
(NEW2) 139 % info vars sim::* # no variable?
(NEW2) 140 % namespace eval sim {set ptr "A string"}
A string
(NEW2) 141 % info vars sim::*
::sim::ptr # but this one exists
(NEW2) 142 % namespace eval sim {set str2 "A string"}
A string
(NEW2) 143 % info vars sim::*
::sim::ptr ::sim::str2 # as does this one?

At this point I figured that maybe just the firstly created was
ignored and tried again

(NEW2) 144 % namespace eval sim {set str [dict create i 0 j 0]}
i 0 j 0
(NEW2) 145 % info vars sim::*
::sim::ptr ::sim::str2 # no, 'str' just isn't created
(NEW2) 146 % namespace eval sim {set d [dict create i 0 j 0]}
i 0 j 0 # same result as in '144'
(NEW2) 147 % info vars sim::*
::sim::d ::sim::ptr ::sim::str2 # but this variable exists (NEW2) 148 % set sim::d
i 0 j 0 # and has the expected value
(NEW2) 149 %

Probably I have a blind spot. I would welcome some feedback.
Thank you
Helmut
Oh, this on Windows 10 64 under Tcl8.6.10

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 2023-06-11, Helmut Giese wrote:

sorry if this is something obvious,

If my guess is correct, it kinda is, and I've stumbled on
exact same problem before.

but I don't know if I can trust my eyes. I have been sweating
several hours over the "problem" to get a command executed inside
a namespace - only to find out, that the cause seems to be
something totally different.

(NEW2) 137 % namespace eval sim {}
(NEW2) 138 % namespace eval sim {set str "A string"}
A string
(NEW2) 139 % info vars sim::* # no variable?

Do you perchance have a global str variable here? If so,
even though wrapped in namespace eval, 'str' above refers
to that /global/ variable.

(NEW2) 140 % namespace eval sim {set ptr "A string"}
A string
(NEW2) 141 % info vars sim::*
::sim::ptr # but this one exists

Conversely, as there're no ptr variable in the upper scope
(presumably), set ptr here creates a namespace variable.

The solution is to always use 'variable' within namespace eval
before referencing a namespace variable. Consider, e. g.:

set foo 0 ; namespace eval foo { variable bar ; set foo 1 ; set bar 2 ; }
puts [ info var foo::* ]
## results in:
::foo::bar

--
FSF associate member #7257 http://am-1.org/~ivan/

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Hello Ivan,
thank you so much, you are totally right: I had a variable 'str' - as
I practically always have.
Once the mystery solved it looks so obvious - but without your help I
would have looked for ages to solve it.
Thanks again
Helmut

On 2023-06-11, Helmut Giese wrote:

sorry if this is something obvious,

If my guess is correct, it kinda is, and I've stumbled on
exact same problem before.

but I don't know if I can trust my eyes. I have been sweating
several hours over the "problem" to get a command executed inside
a namespace - only to find out, that the cause seems to be
something totally different.

(NEW2) 137 % namespace eval sim {}
(NEW2) 138 % namespace eval sim {set str "A string"}
A string
(NEW2) 139 % info vars sim::* # no variable?

Do you perchance have a global str variable here? If so,
even though wrapped in namespace eval, 'str' above refers
to that /global/ variable.

(NEW2) 140 % namespace eval sim {set ptr "A string"}
A string
(NEW2) 141 % info vars sim::*
::sim::ptr # but this one exists

Conversely, as there're no ptr variable in the upper scope
(presumably), set ptr here creates a namespace variable.

The solution is to always use 'variable' within namespace eval
before referencing a namespace variable. Consider, e. g.:

set foo 0 ; namespace eval foo { variable bar ; set foo 1 ; set bar 2 ; } >puts [ info var foo::* ]
## results in:
::foo::bar

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 6/11/2023 10:06 AM, Helmut Giese wrote:

Hello out there,
sorry if this is something obvious, but I don't know if I can trust my
eyes. I have been sweating several hours over the "problem" to get a
command executed inside a namespace - only to find out, that the cause
seems to be something totally different.
Consider this screen dump (from within TKCon):
(NEW2) 136 % catch {namespace delete sim}

0
(NEW2) 137 % namespace eval sim {}
(NEW2) 138 % namespace eval sim {set str "A string"}
A string
(NEW2) 139 % info vars sim::* # no variable?
(NEW2) 140 % namespace eval sim {set ptr "A string"}
A string
(NEW2) 141 % info vars sim::*
::sim::ptr # but this one exists
(NEW2) 142 % namespace eval sim {set str2 "A string"}
A string
(NEW2) 143 % info vars sim::*
::sim::ptr ::sim::str2 # as does this one?

At this point I figured that maybe just the firstly created was
ignored and tried again

(NEW2) 144 % namespace eval sim {set str [dict create i 0 j 0]}
i 0 j 0
(NEW2) 145 % info vars sim::*
::sim::ptr ::sim::str2 # no, 'str' just isn't created
(NEW2) 146 % namespace eval sim {set d [dict create i 0 j 0]}
i 0 j 0 # same result as in '144'
(NEW2) 147 % info vars sim::*
::sim::d ::sim::ptr ::sim::str2 # but this variable exists
(NEW2) 148 % set sim::d
i 0 j 0 # and has the expected value
(NEW2) 149 %

Probably I have a blind spot. I would welcome some feedback.
Thank you
Helmut
Oh, this on Windows 10 64 under Tcl8.6.10

on windows 10, tcl 8.6.9, from console

()1 % catch {namespace delete sim}
1
()2 % namespace eval sim {}
()3 % namespace eval sim {set str "A string"}
A string
()4 % info vars sim::*
::sim::str
()5 % namespace eval sim {set ptr "A string"}
A string
()6 % info vars sim::*
::sim::str ::sim::ptr
()7 % namespace eval sim {set str2 "A string"}
A string
()8 % info vars sim::*
::sim::str ::sim::ptr ::sim::str2
()9 % namespace eval sim {set str [dict create i 0 j 0]}
i 0 j 0
()10 % info vars sim::*
::sim::str ::sim::ptr ::sim::str2
()11 % namespace eval sim {set d [dict create i 0 j 0]}
i 0 j 0
()12 % info vars sim::*
::sim::d ::sim::str ::sim::ptr ::sim::str2
()13 % set sim::d
i 0 j 0
()14 % info pa
8.6.9


I suggest you try with windows console, and try from a newly run program,
could be something funny happened in the previous 135 commands, or maybe something in tkcon.

Here's the text as I pasted it in, i had to edit out one of your comments

catch {namespace delete sim}
namespace eval sim {}
namespace eval sim {set str "A string"}
info vars sim::*
namespace eval sim {set ptr "A string"}
info vars sim::*
namespace eval sim {set str2 "A string"}
info vars sim::*
namespace eval sim {set str [dict create i 0 j 0]}
info vars sim::*
namespace eval sim {set d [dict create i 0 j 0]}
info vars sim::*
set sim::d

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

Hello et99,
thanks for your effort, without Ivan's answer I would have tried your
way and it would have shed some light on this problem - but Ivan's
idea was spot on: a 'variable' statement for every newly created
variable - and the problem was gone.
Thanks again
Helmut

On 6/11/2023 10:06 AM, Helmut Giese wrote:

Hello out there,
sorry if this is something obvious, but I don't know if I can trust my
eyes. I have been sweating several hours over the "problem" to get a
command executed inside a namespace - only to find out, that the cause
seems to be something totally different.
Consider this screen dump (from within TKCon):
(NEW2) 136 % catch {namespace delete sim}

0
(NEW2) 137 % namespace eval sim {}
(NEW2) 138 % namespace eval sim {set str "A string"}
A string
(NEW2) 139 % info vars sim::* # no variable?
(NEW2) 140 % namespace eval sim {set ptr "A string"}
A string
(NEW2) 141 % info vars sim::*
::sim::ptr # but this one exists
(NEW2) 142 % namespace eval sim {set str2 "A string"}
A string
(NEW2) 143 % info vars sim::*
::sim::ptr ::sim::str2 # as does this one?

At this point I figured that maybe just the firstly created was
ignored and tried again

(NEW2) 144 % namespace eval sim {set str [dict create i 0 j 0]}
i 0 j 0
(NEW2) 145 % info vars sim::*
::sim::ptr ::sim::str2 # no, 'str' just isn't created
(NEW2) 146 % namespace eval sim {set d [dict create i 0 j 0]}
i 0 j 0 # same result as in '144'
(NEW2) 147 % info vars sim::*
::sim::d ::sim::ptr ::sim::str2 # but this variable exists
(NEW2) 148 % set sim::d
i 0 j 0 # and has the expected value
(NEW2) 149 %

Probably I have a blind spot. I would welcome some feedback.
Thank you
Helmut
Oh, this on Windows 10 64 under Tcl8.6.10

on windows 10, tcl 8.6.9, from console

()1 % catch {namespace delete sim}
1
()2 % namespace eval sim {}
()3 % namespace eval sim {set str "A string"}
A string
()4 % info vars sim::*
::sim::str
()5 % namespace eval sim {set ptr "A string"}
A string
()6 % info vars sim::*
::sim::str ::sim::ptr
()7 % namespace eval sim {set str2 "A string"}
A string
()8 % info vars sim::*
::sim::str ::sim::ptr ::sim::str2
()9 % namespace eval sim {set str [dict create i 0 j 0]}
i 0 j 0
()10 % info vars sim::*
::sim::str ::sim::ptr ::sim::str2
()11 % namespace eval sim {set d [dict create i 0 j 0]}
i 0 j 0
()12 % info vars sim::*
::sim::d ::sim::str ::sim::ptr ::sim::str2
()13 % set sim::d
i 0 j 0
()14 % info pa
8.6.9

I suggest you try with windows console, and try from a newly run program, >could be something funny happened in the previous 135 commands, or maybe >something in tkcon.

Here's the text as I pasted it in, i had to edit out one of your comments

catch {namespace delete sim}
namespace eval sim {}
namespace eval sim {set str "A string"}
info vars sim::*
namespace eval sim {set ptr "A string"}
info vars sim::*
namespace eval sim {set str2 "A string"}
info vars sim::*
namespace eval sim {set str [dict create i 0 j 0]}
info vars sim::*
namespace eval sim {set d [dict create i 0 j 0]}
info vars sim::*
set sim::d

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 11/06/2023 19:53, Ivan Shmakov wrote:

On 2023-06-11, Helmut Giese wrote:

> (NEW2) 139 % info vars sim::* # no variable?

Do you perchance have a global str variable here? If so,
even though wrapped in namespace eval, 'str' above refers
to that /global/ variable.

To be fixed in Tcl 9.0: https://core.tcl-lang.org/tips/doc/trunk/tip/278.md


Schelte.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 6/11/23 12:06, Helmut Giese wrote:

Hello out there,
sorry if this is something obvious, but I don't know if I can trust my
eyes. I have been sweating several hours over the "problem" to get a
command executed inside a namespace - only to find out, that the cause
seems to be something totally different.
Consider this screen dump (from within TKCon):
(NEW2) 136 % catch {namespace delete sim}

0
(NEW2) 137 % namespace eval sim {}
(NEW2) 138 % namespace eval sim {set str "A string"}
A string
(NEW2) 139 % info vars sim::* # no variable?
(NEW2) 140 % namespace eval sim {set ptr "A string"}
A string
(NEW2) 141 % info vars sim::*
::sim::ptr # but this one exists
(NEW2) 142 % namespace eval sim {set str2 "A string"}
A string
(NEW2) 143 % info vars sim::*
::sim::ptr ::sim::str2 # as does this one?

At this point I figured that maybe just the firstly created was
ignored and tried again

(NEW2) 144 % namespace eval sim {set str [dict create i 0 j 0]}
i 0 j 0
(NEW2) 145 % info vars sim::*
::sim::ptr ::sim::str2 # no, 'str' just isn't created
(NEW2) 146 % namespace eval sim {set d [dict create i 0 j 0]}
i 0 j 0 # same result as in '144'
(NEW2) 147 % info vars sim::*
::sim::d ::sim::ptr ::sim::str2 # but this variable exists
(NEW2) 148 % set sim::d
i 0 j 0 # and has the expected value
(NEW2) 149 %

Probably I have a blind spot. I would welcome some feedback.
Thank you
Helmut
Oh, this on Windows 10 64 under Tcl8.6.10

Because the variable str existed in the global namespace before line 138!!

Either unset it or use the variable command instead of the set command
to set the value.

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)