| Ticket UUID: | 3047235 | |||
| Title: | non-bc'ed catch and tailcall | |||
| Type: | Bug | Version: | obsolete: 8.6b1.1 | |
| Submitter: | msofer | Created on: | 2010-08-17 20:13:22 | |
| Subsystem: | 60. NRE and coroutines | Assigned To: | msofer | |
| Priority: | 9 Immediate | Severity: | ||
| Status: | Closed | Last Modified: | 2010-08-19 05:34:10 | |
| Resolution: | Fixed | Closed By: | msofer | |
| Closed on: | 2010-08-18 22:34:10 | |||
| Description: |
[Moved from 3046594]
You cannot [tailcall] out of a catch-environment. Bytecompiled [catch] is working properly and tested (tailcall-12.3a/b), non-bytecompiled [catch] is neither and segfaults:
% proc foo {} {set catch catch; $catch {tailcall concat}}
% foo
Program received signal SIGSEGV, Segmentation fault.
| |||
| User Comments: |
msofer added on 2010-08-19 05:34:10:
allow_comments - 1 New redesign committed: [tailcall] errors out early, the command info is logged by the usual means. msofer added on 2010-08-19 01:18:46: allow_comments - 0 Missing Tcl_LogCommandInfo ... back to the drawing board msofer added on 2010-08-19 00:51:25: Fishy:
% proc foo {} {set catch catch; list [$catch {tailcall concat} msg] $msg}
% foo
1 {tailcall called from within a catch environment}
% set errorInfo
can't read "errorInfo": no such variable
dgp added on 2010-08-18 23:56:16:
confirm no crash, but either it or my expectation are
still not right:
% proc foo {} {set catch catch; $catch {tailcall concat}}
% foo
1
% set errorInfo
can't read "errorInfo": no such variable
What error is [catch] catching here? And
how can we programmatically get a hold of it?
msofer added on 2010-08-18 23:40:41: Tailcall should never be able to escape a catch context. Please try it (and commit a test?). ferrieux added on 2010-08-18 23:32:37: Now does it mean that a 'tailcall tailcall tailcall' could pop back 3 levels under the nose of a [catch] in the caller's context, without being caught ? msofer added on 2010-08-18 22:47:54: allow_comments - 1 msofer added on 2010-08-18 22:47:53: Fixed in HEAD by redsigning (and simplifying) tailcall msofer added on 2010-08-18 06:46:02: File Added - 383752: diff msofer added on 2010-08-18 06:45:29: Attached patch fixes this issue, but new test tailcall-12.3c reveal something strange is still amiss. | |||
Attachments:
- diff [download] added by msofer on 2010-08-18 06:46:02. [details]