Apache OpenOffice (AOO) Bugzilla – Issue 69836
CurDir() function is not Unicode-compliant
Last modified: 2013-02-24 21:07:33 UTC
Curdir returns garbage if current directory path contains non-ascii characters. Attached is a patch which converts result according current gsl_getSystemTextEncoding. Can you investigate this patch and apply if it is OK?
Created attachment 39356 [details] curdir diff
2.1 STARTED
FIXED
Reopen
ab->rail: I realised too late that your code does not work at least for Windows, as the _getdcwd system call is used that does not support Unicode. Anyway this is old system dependend code and should be replaced by system inde- pendend vos API calls. Maybe you want to provide a patch for this? -> INVALID
rail->ab: can you provide your test case which not pass? I use very simple one and it works for me. 1. Run "cmd" 2. mkdir c:\теÑÑ‚ 3. cd c:\теÑÑ‚ 4. run OOo via cmd (c:\Program Files\...) 5. Add new module/macro: Sub Test MsgBox CurDir() End Sub Box shows proper value.
ab->rail: I can't find my test case any more. But even without it there is the problem that documentation for _getdcwd says that this is not the right command for Unicode. Anyway I reopen the tast and will have another look. This could take some time as currently I'm very busy, Reopen
OOo 2.x, STARTED
Target => 2.3
Still planned for OOo 2.3, but currently no time available for checking again
ab->rail: I had another look now, but still can't verify the patch. I don't understand how I can start OOo in a way that makes c:\теÑÑ‚ the current directory. As already stated in i69835 I cannot access directory containing unicode characters from command prompt as there only ???? is displayed. cd c:\теÑÑ‚ simply is not possible on my system (Windows Server 2003 SP1). Any idea?
I have found no way to test the patch and set it back to INVALID. If a test case will become available later, the task can be reopened.
CLOSED