From: "Tom Demmer" <DEMMER AT brain1 DOT lstm DOT ruhr-uni-bochum DOT de>
Organization:  Lehrstuhl Stroemungsmechanik, RUB
To: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>, djgpp-workers AT delorie DOT com
Date:          Thu, 8 Aug 1996 08:54:19 GMT-1
Subject:       Re: UNC for DJGPP?
Reply-to: Demmer AT LStM DOT Ruhr-Uni-Bochum DOT De
Message-ID: <4C794433282@brain1.lstm.ruhr-uni-bochum.de>

> 
> On Wed, 7 Aug 1996, Tom Demmer wrote:
> 
> > I had a little spare time the last days and wondered how much work it 
> > might be to teach djgpp a bit UNC. Surprinsingly it was pretty easy, 
> > a bit hacking in fixpath(), a bit in stat() and a lot of things seem 
> > to work out fine. At least with Novell VLMs.
> 
> Are you sure this will work on other redirectors?  What about 
> `_put_path'?  AFAIK, it weeds out multiple slashes, and it might 
> therefore convert "//machine/vol/dir" into "/machine/vol/dir".  So if 
> some application runs "\\machine\vol\dir" through the usual loop of 
> converting the slashes, they will be in trouble, right?  And `stat' is 
> quite entrenched in assuming it isn't called with a UNC; are you sure you 
> looked at the latest sources of `stat' (in the alpha 960725 snapshot)?
It is the one you sent me that corrects the NETX Bug for 
stat("i:/subdir/."); Where can I find these alpha sources?

The rule for having (or recognizing) a UNC is that it _MUST_ start 
with two backslashes. This complies with the samba deamon on Linux.
DOS does not recognize "//machine/vol/dir", but it recognizes 
"\\machine/vol/dir" and "\\machine/vol\dir/subdir\subdir"

DOS itself seems to be pretty flexible, most things are broken by 
command.com.

For the other redirectos: This is quite easy to check, I guess if
_truename("[NetDrv_Letter]:")  returns a UNC, it should work. But I 
do not have more redirectors here right now. But I am setting up a 
samba deamon for NetBIOS  on another network, so I'll be able to 
check that soon. 

I tested this stuff so far with NetWare VLM and XFS, both work fine. 
Netware drives can be accessed with UNCs, XFS drives cannot, but 
_truename("e:bin") just returns "e:\bin". stat() still recognizes 
that e: is not a local drive and invents inodes. 
What i have not yet tried is system("\\\\machine/vol/dir/prog") and
the spawn* and exec* calls. 

One more thing to consider is splitpath(). I think it is a good idea 
to put the machine name into the drive slot, but for that  I need to 
know the maximum allowed length for the name. Some nice guy on the 
german novell list will know that for sure. 

> 
> > Do you think this is 
> > something to incorporate into the next release? It would ease the 
> > pains of a poor beaten NetWare administrator when he can forget about 
> > things like ``drive mapping'' and such.
> 
> I'm not sure, but if it doesn't break other things, I don't mind.  

Not breaking other things has highest priority.

> Why  won't you ask djgpp-workers and see what other people think?  

They'll get a copy of this. 

> The issue  of forward and backslashes in a UNC should be resolved 
> for this to work, IMHO.
  
Sure.

Ciao
Tom

  ******************************************************************
  *  Thomas Demmer                     * Phone : +49 234 700 6434  *
  *  Universitaetsstr. 150             * Fax   : +49 234 709 4162  *
  *  Lehrstuhl fuer Stroemungsmechanik * Fax/Voice Box:            *
  *  D-44780 Bochum                    *   +49 2561 91371 2056     *
  ******************************************************************
  *  Email: demmer AT LStM DOT Ruhr-Uni-Bochum DOT De                         *
  *  WWW:   http://www.lstm.ruhr-uni-bochum.de/~demmer             *
  ******************************************************************

    Muysenberg's Law: There's an easier way to do anything.