Ticket UUID: | 1386377 | |||
Title: | clock scan [clock format [clock seconds] -gmt 1] | |||
Type: | Bug | Version: | obsolete: 8.5a4 | |
Submitter: | nobody | Created on: | 2005-12-20 18:05:52 | |
Subsystem: | 06. Time Measurement | Assigned To: | kennykb | |
Priority: | 5 Medium | Severity: | ||
Status: | Open | Last Modified: | 2006-07-31 10:36:51 | |
Resolution: | None | Closed By: | ||
Closed on: | ||||
Description: |
clock scan [clock format [clock seconds] -gmt 1] fails with: unable to convert date-time string "Tue Dec 20 17:41:41 +0000 2005" while evaluating {clock scan [clock format [clock seconds] -gmt 1]} The default format has changed from 8.4 to current CVS. tclsh8.4 [~]clock format [clock seconds] Tue Dec 20 01:00:34 PM EST 2005 tclsh8.5 [~]clock format [clock seconds] Tue Dec 20 13:02:10 -0500 2005 Not sure if this was desired, but a trivial patch (attached) restores the old behaviour. | |||
User Comments: |
kennykb added on 2006-07-31 10:36:51:
Logged In: YES user_id=99768 In the 8.5 HEAD, I've put [clock format] back to using the timezone name rather than the value. We really, really need to make positive as well as negative timezones work in free-form [clock scan] - but that's a tremendous amount of work. There are simply too many ambiguities for the YACC parser to do The Right Thing. I'm leaving this open because I have some other ideas (short of resorting to fixed-format [clock scan]) for fixing this in 8.5. cmay added on 2005-12-21 01:15:40: Logged In: YES user_id=191458 But maybe [clock scan] should be able to handle more formats ? nobody added on 2005-12-21 01:05:52: File Added - 160560: tcl_clock.patch.txt |
Attachments:
- tcl_clock.patch.txt [download] added by nobody on 2005-12-21 01:05:52. [details]