|
2009-04-03
| ||
| 02:51 | • Ticket [2433843fff] return options can leak to next bytecode status still Op... artifact: 320738a7bf user: dgp | |
|
2008-12-17
| ||
| 00:00 | • Ticket [2433843fff]: 1 change artifact: db77912805 user: dgp | |
|
2008-12-16
| ||
| 16:41 | • New ticket [2433843fff]. artifact: 49f590a984 user: dgp | |
| Ticket UUID: | 2433843 | ||
| Title: | return options can leak to next bytecode | ||
| Type: | Bug | Created on: | 2008-12-16 16:41:41 |
| Submitter: | dgp | Assigned to: | dgp |
| Subsystem: | 47. Bytecode Compiler | Severity: | |
| Priority: | 5 Medium | Last modified: | 2009-04-03 02:51:19 |
| Status: | Open | Closed by: | |
| Resolution: | None | Closed on: | |
| Version: | obsolete: 8.5.5 | ||
| Description: | ||||
% proc demo {} {
return -level 0 -code ok -leaky option
set x
}
% catch demo m o
1
% set o
-leaky option -code 1 -level 0 -errorcode NONE -errorinfo {can't read "x": no such variable
while executing
"set x"
(procedure "demo" line 3)
invoked from within
"demo"} -errorline 1
% info patch
8.5.5
Problem appears to be that all commands
returning TCL_OK bypass any (emulation of)
Tcl_ResetResult() on the theory that results
are managed on the execution stack so long
as bytecode is in control. This means
resetting of the return options dict gets
skipped.
| ||||
| User Comments: | ||||
dgp added on 2009-04-03 02:51:19:
see also 2007198 | ||||
