X-Authentication-Warning: new-smtp1.ihug.com.au: Host p46-tnt4.syd.ihug.com.au [203.173.134.46] claimed to be acceleron
Message-ID: <017401c11e7a$b3493f90$0a02a8c0@acceleron>
From: "Andrew Cottrell" <acottrel AT ihug DOT com DOT au>
To: "Andrew Cottrell" <acottrel AT ihug DOT com DOT au>,
        "Eli Zaretskii" <eliz AT is DOT elta DOT co DOT il>,
        "Charles Sandmann" <sandmann AT clio DOT rice DOT edu>
Cc: <djgpp-workers AT delorie DOT com>
Subject: Re: Windows 2000 /dev/null permission query
Date: Mon, 6 Aug 2001 23:21:26 +1000
MIME-Version: 1.0
Content-Type: text/plain;
	charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Reply-To: djgpp-workers AT delorie DOT com

> >
> > > Conclusion - LFN support in W2K is breaking things (like truncate,
> > > utime, get_dev_info).  Handles opened with the LFN calls are not being
> > > treated the same as those opened with the old DOS APIs.
> >
> > Does function 7160/CL=01 (get short 8+3 name for a long file name)
> > work reliably on W2K?  If so, we could selectively modify functions
> > like utime that open files and get trouble on W2K, to convert the long
> > name to a short name and then use non-LFN functions to open the file.
> > Only on W2K/XP, of course.
> Added a highly modified version of _truename in to my touch example and
only
> called the new function. Under Win2K the call worked on my two examples.
The
> test and output was:
>
>  DJGPP_204 D:\dj204\contrib\touch>dir ac*.txt
> 06/08/2001  10:31p                  22 acnewfilename.txt
> 06/08/2001  10:39p                   8 acnewfilename2.txt
>
> DJGPP_204 D:\dj204\contrib\touch>touch
> True name of acnewfilename.txt is D:\dj204\contrib\touch\ACNEWF~1.TXT
> True name of acnewfilename2.txt is D:\dj204\contrib\touch\ACNEWF~2.TXT
>
> DJGPP_204 D:\dj204\contrib\touch>dir acnewf~1.txt
> 06/08/2001  10:31p                  22 acnewfilename.txt
>
> DJGPP_204 D:\dj204\contrib\touch>dir acnewf~2.txt
> 06/08/2001  10:39p                   8 acnewfilename2.txt
>
> The function also worked fine on Win98.
> > We could also do this universally (on W2K/XP), in all functions that
> > open files, but for that we will have first to see if opening a file
> > with a legacy DOS function disables some LFN features when we pass a
> > handle from a non-LFN open.  For example, do functions 5704 and 5705
> > work on a handle open by a non-LFN DOS call?
Checked the code and it looked so easy to test so....
If in _open.c I use the SFN code ah = 0x3d, al = oflag or the 0x6C00 call
then touch updates the time with the 5701 DPMI int and the int86 call.

5704 / 5705 and 5706/5707 appear to work with the mod in _open, but I need
to check the files properties to ensure that they work.

I need to double check tomorrow night to ensure I have not made any mistakes
and add extra debugging to make the test easier to check.

> > Another idea might be to use `dup' to duplicate a handle from an LFN
> > open call, then see if that handle has the same problems with function
> > 5701 as the original handle.  (Since `dup' doesn't have an LFN analog,
> > maybe the duplicated handle is created as a non-LFN one, who knows?)
>
 Not tested. Tomorrow