|
2026-03-25
| ||
| 04:11 | Merge 9.0. Bug [b0682c3c24]. Windows Glob error with drive in pattern check-in: 6cd2936408 user: apnadkarni tags: trunk, main | |
| 03:31 | • Closed ticket [b0682c3c24]: glob error plus 6 other changes artifact: ca18c9cc2d user: apnadkarni | |
| 03:29 | Bug [b0682c3c24]. Windows Glob error with drive in pattern check-in: 874e86e528 user: apnadkarni tags: core-9-0-branch | |
|
2026-03-18
| ||
| 16:20 | • Ticket [b0682c3c24] glob error status still Open with 3 other changes artifact: 19d58ec013 user: apnadkarni | |
| 16:04 | • Ticket [b0682c3c24]: 4 changes artifact: 30f75b7c06 user: apnadkarni | |
| 11:38 | Start on bug [b0682c3c24]. glob errors on Windows check-in: 52f35167af user: apnadkarni tags: bug-b0682c3c24 | |
|
2025-11-14
| ||
| 13:22 | • Ticket [b0682c3c24] glob error status still Open with 3 other changes artifact: aa2a0c3ed2 user: anonymous | |
| 07:48 | • Ticket [b0682c3c24]: 3 changes artifact: 58497f4248 user: anonymous | |
|
2025-10-07
| ||
| 15:31 | • Ticket [b0682c3c24]: 3 changes artifact: 745924f882 user: dkf | |
|
2025-09-25
| ||
| 16:31 | • Ticket [b0682c3c24]: 3 changes artifact: a4c316b177 user: juliannoble2 | |
| 12:52 | • Ticket [b0682c3c24]: 3 changes artifact: e3e8fccf7d user: anonymous | |
|
2025-07-21
| ||
| 19:15 | • Ticket [b0682c3c24]: 3 changes artifact: a32b0ae555 user: sebres | |
|
2025-06-10
| ||
| 09:40 | • New ticket [b0682c3c24]. artifact: ea513ab823 user: anonymous | |
| Ticket UUID: | b0682c3c2497e8f27f18e91686031dda055ed19c | ||
| Title: | glob error | ||
| Type: | Bug | Created on: | 2025-06-10 09:40:06 |
| Submitter: | anonymous | Assigned to: | nobody |
| Subsystem: | 37. File System | Severity: | Important |
| Priority: | 5 Medium | Last modified: | 2026-03-25 03:31:11 |
| Status: | Closed | Closed by: | apnadkarni |
| Resolution: | Fixed | Closed on: | 2026-03-25 03:31:11 |
| Version: | 9.0.0 9.0.1 | ||
| Description: | ||||
I have the following strange behaviour under Windows 10 with tcl_patchLevel=9.0.0 (and tcl_patchLevel=9.0.1 from BAWT)
(bin) 1 % set tcl_patchLevel
9.0.0
(bin) 2 % glob Z:/Basistests/200*
Z:/Basistests/2000_00_00 {Z:/Basistests/2000_00_00 - Test}
(bin) 3 % glob -types d Z:/Basistests/200*
Z:/Basistests/2000_00_00 {Z:/Basistests/2000_00_00 - Test}
(bin) 4 % glob -types d -directory Z:/Basistests -tails Z:/Basistests 200*
Z:Basistests 2000_00_00 {2000_00_00 - Test}
(bin) 5 % glob -types d -directory Z:/Basistests -tails Z:/Basistests Z:/Basistests/200*
couldn't read directory "Z:/Basistests/Z:Basistests/200*": not a directory
Under tcl8 I got:
(ABS) 1 % glob -types d -directory Z:/Basistests -tails Z:/Basistests 200*
2000_00_00 {2000_00_00 - Test}
(ABS) 2 % glob -types d -directory Z:/Basistests -tails Z:/Basistests Z:/Basistests/200*
no files matched glob patterns "Z:/Basistests Z:/Basistests/200*"
The additional string "Z:Basistests" on comcd mand 4 is wrong.
Even more perplexing is the following (under Tcl/Tk 9.0.0):
(ABS) 1 % glob -types d -directory Z:/Basistests -tails Z:/Basistests 200*
Z:Basistests 2000_00_00 {2000_00_00 - Test}
(ABS) 2 % cd Z:Basistests
(Basistests) 3 % pwd
Z:/Basistests
(Basistests) 4 % cd C:
() 5 % cd Z:Basistests
(Basistests) 6 % cd 2000_00_00
(2000_00_00) 7 % cd Z:Basistests
couldn't change working directory to "Z:Basistests": no such file or directory
(2000_00_00) 8 %
| ||||
| User Comments: | ||||
sebres added on 2025-07-21 19:15:09:
If I understand it correctly (please provide short "diff", what is expected and what is wrong result), it may be a side effect of bug [61c01e0edb08a9ed], because looks really similar to us. If so it shall be fixed now. Please test again with the fixed version. anonymous added on 2025-09-25 12:52:31:
Still the same under tcl 9.0.2 :( juliannoble2 added on 2025-09-25 16:31:19:
I also see this on a recent 9.0.2 build Paths such as Z:Basistests are volume-relative Tcl doesn't support the per-volume cwd - there are long-standing bugs in this area when changing drives in Tcl as compared to what happens in cmd.exe or powershell. If you cd to Z:/Basistests for command 4 - you will see the Z:basistests entry is no longer present in the result. for command 5 - the 'couldn't read directory' error is not raised. glob with the -directory option, and no specific reference to volume-relative paths shouldn't change output based on the cwd. (and shouldn't return any volume relative paths) I've argued in a previous ticket that it may be worth considering removing volume-relative path support. I'd really prefer it was supported - but there are still footguns regarding the cd command and comparing glob output to external tools/commands view of things due to the occasional difference in ideas of what is the actual current working directory for a drive. The example given above of cd c: & cd Z:Basistests can be a problem with tcl8 too - as it can depend on what directory you were in when starting tclsh. Perhaps in this case we can at least make glob handle things better, even if a full fix-up of the volume-relative path problems isn't something anyone is able to delve into now. anonymous added on 2025-11-14 07:48:15:
It seems the error is gone in tcl9.0.3. anonymous added on 2025-11-14 13:22:37:
Sorry, the error gone in tcl9.0.3 is not true :( The strange Z:Basistests still exists. apnadkarni added on 2026-03-18 16:04:36:
Potential fix in the bug-b0682c3c24 branch. But see below.
which seems reasonable to me. The cd issues in the original ticket I didn't quite understand as it seems to be it depends on what the "current directory" on the other drive happened to be. In the existing implementation, glob correctly parses the As a fix, it seems to me that
I therefore consider this as a bug fix and not a change to defined or expected Opinions? apnadkarni added on 2026-03-18 16:20:36:
Regarding the cd issue, I've logged [bca391ab51] as a separate issue. May be that is what the ticket was referring to.
apnadkarni added on 2026-03-25 03:31:11:
Fixed in [874e86e528]. | ||||
