| Ticket UUID: | 1274916 | ||
| Title: | upvar and qualified varNames | ||
| Type: | Bug | Created on: | 2005-08-28 01:34:22 |
| Submitter: | msofer | Assigned to: | msofer |
| Subsystem: | 07. Variables | Severity: | |
| Priority: | 7 High | Last modified: | 2009-05-01 08:21:38 |
| Status: | Open | Closed by: | |
| Resolution: | None | Closed on: | |
| Version: | None | ||
| Description: | ||||
The behaviour of
upvar otherVar myVar
is wrong when myVar is a qualified name, as it does not
follow Tcl's (insane?) variable resolution rules: it
skips the global lookup upon failure of the local one.
Witness:
% set x 1
1
% namespace eval a {upvar x b::x}
can't create "b::x": parent namespace doesn't exist
% namespace eval b {}
% namespace eval a {upvar x b::x}
can't create "b::x": parent namespace doesn't exist
% namespace eval b {variable x 2}
% namespace eval a {upvar x b::x}
can't create "b::x": parent namespace doesn't exist
% namespace eval a {set b::x}
2
This worked as expected in Tcl8.3.4, and is broken as
described above in 8.4.7 and HEAD. Breakage was
probably introduced at some point in the 8.4 line
(LOOKUP_FOR_UPVAR).
| ||||
| User Comments: | ||||
msofer added on 2005-08-28 08:51:53:
Logged In: YES user_id=148712 Note that this behaviour was introduced as a fix for Bug #696893. Avoiding the global lookup appears not to have been a correct reconstruction of Tcl8.3's lookup rules. msofer added on 2005-08-28 19:37:39:
Logged In: YES
user_id=148712
The problem is that [upvar] has different rules for creative
writing in the not-fully-qualified case. In the example above
namespace eval a {upvar x b::x}
the new variable should refer (in order) to:
(1) ::a::b::x if the namespace ::a::b exists
(2) ::b::x if the namespace ::b exists
Note that this refers to the existence of namespaces; the
usual resolution rules refer to the existence of the
variables themselves, ie for [namespace eval a {set b::x 1}]
the reference is to:
(1) ::a::b::x if the variable ::a::b::x exists
(2) ::b::x if the variable ::b::x exists
(3) ::a::b::x otherwise (even if ::b exists but ::a::b
does not)
msofer added on 2005-08-28 19:47:17:
Logged In: YES user_id=148712 The question now is if this behaviour in 8.3 is intended, or just an implementation side effect. Ths docs do not even mention the possibility of creating anything but a proc-local variable via upvar. | ||||
