| Ticket UUID: | 219142 | |||
| Title: | bug in bgerror: should use Tcl_GetCommandInfo | |||
| Type: | Bug | Version: | obsolete: 8.0.3 | |
| Submitter: | nobody | Created on: | 2000-10-26 05:02:16 | |
| Subsystem: | 02. Event Loops | Assigned To: | msofer | |
| Priority: | 6 | Severity: | ||
| Status: | Closed | Last Modified: | 2001-12-11 03:31:49 | |
| Resolution: | Fixed | Closed By: | msofer | |
| Closed on: | 2001-12-10 20:31:48 | |||
| Description: |
OriginalBugID: 750 Bug Version: 8.0.3 SubmitDate: '1998-10-22' LastModified: '2000-01-23' Severity: MED Status: Assigned Submitter: admin ChangedBy: hobbs RelatedBugIDs: 3998 2577 2136 OS: All FixedDate: '2000-10-25' FixedInVersion: NA ClosedDate: '2000-10-25' Name: Paul Duffin Tcl_BackgroundError evaluates bgerror in the global context. If that returns an error then it generates the message. To determine whether bgerror is a valid command it compares the interp result with a particular string and if it matches then bgerror is not defined, else bgerror generated an error. In the first case it just writes out what the original error was, in the second case it writes out both the original error and the new error (as above). Unfortunately the string it tests against is not invalid command name "bgerror" but "bgerror" is an invalid command name or ambiguous abbreviation which I presume used to be the error returned when a command did not exist but the error has changed. It should really use Tcl_GetCommandInfo or something similar to check whether bgerror is defined. tclEvent.c:HandleBgErrors is where the problem is, but this has other problems that need resolving (see related reports) at the same time. -- 01/23/2000 hobbs | |||
| User Comments: |
msofer added on 2001-12-11 03:31:49:
File Added - 14469: bgerror.patch msofer added on 2001-12-11 03:31:48: Logged In: YES user_id=148712 Single loc fix and test committed. kennykb added on 2001-12-11 03:20:24: Logged In: YES user_id=99768 Donal's not entirely right here. The stack trace of the original error gets lost. dkf added on 2000-12-03 01:54:38: OTOH, it is not completely disastrous though. All you get is a complaint about the lack of a bgerror *and* a report of the original message as well... dkf added on 2000-11-19 17:48:52: Ought to investigate further, but the problem will only show up when you've not got a bgerror command defined, and that's usually bad practise when you're using the event loop. Still, fixing this really ought to be a priority. | |||
Attachments:
- bgerror.patch [download] added by msofer on 2001-12-11 03:31:48. [details]