Ticket UUID: | 1055836 | |||
Title: | TclThread_Init called twice? | |||
Type: | Bug | Version: | obsolete: 8.5a2 | |
Submitter: | davygrvy | Created on: | 2004-10-28 03:08:04 | |
Subsystem: | 34. tcltest Package | Assigned To: | davygrvy | |
Priority: | 5 Medium | Severity: | ||
Status: | Closed | Last Modified: | 2004-10-28 13:25:53 | |
Resolution: | Fixed | Closed By: | davygrvy | |
Closed on: | 2004-10-28 06:25:53 | |||
Description: |
from Tcl_AppInit() of win/tclAppInit.c, Tcltest_Init is called and from within that, it calls TclThread_Init (line 710 of generic/tclTest.c), yet Tcl_AppInit calls TclThread_Init a second time (line 164 of win/tclAppInit.c). How is right? Should Tcl_AppInit be responsible or Tcltest_Init for loading the thread package that is part of the test suite? | |||
User Comments: |
davygrvy added on 2004-10-28 13:25:53:
Logged In: YES user_id=7549 changes committed to HEAD. davygrvy added on 2004-10-28 11:49:47: Logged In: YES user_id=7549 Ok, will do. Tcl_AppInit calling TclThread_Init shall get hosed. I once thought the same about the thread package, but Jeff wanted the test code to stay within the core itself. I don't know he feels lately about it. A good goal, though. dgp added on 2004-10-28 11:41:50: Logged In: YES user_id=80530 It's definitely wrong, though harmless, to have it called from both places. Feel free to make whatever temporary fix you prefer. The correct long term fix is to destroy the phony, obsolete "Thread 1.0" provided by TclThread_Init() completely, and simply bundle the real, [load]-able Thread package with the core. davygrvy added on 2004-10-28 10:20:55: File Added - 106783: patch.txt Logged In: YES user_id=7549 Personally, I think the shell should not be responsible as it is good refactoring. See attached patch. |
Attachments:
- patch.txt [download] added by davygrvy on 2004-10-28 10:20:55. [details]