www.delorie.com/archives/browse.cgi   search  
Mail Archives: djgpp/1998/11/03/11:01:36

Message-Id: <199811031559.KAA20728@scrooge.chesapeake.rps.slb.com>
X-Sender: hackbart AT chesapeake DOT rps DOT slb DOT com
X-Mailer: QUALCOMM Windows Eudora Pro Version 4.0.2
Date: Tue, 03 Nov 1998 11:01:45 -0500
To: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
From: Mike Hackbart <hackbart AT chesapeake DOT rps DOT slb DOT com>
Subject: Re: RCS and Network Drives - ENODEV, ENOENT
Cc: djgpp AT delorie DOT com
In-Reply-To: <Pine.SUN.3.91.981103160352.20396B-100000@is>
References: <4 DOT 0 DOT 2 DOT 19981103070733 DOT 00ad3d70 AT chesapeake DOT rps DOT slb DOT com>
Mime-Version: 1.0
Reply-To: djgpp AT delorie DOT com

At 04:12 PM 11/3/98 +0200, Eli Zaretskii wrote:
>
>On Tue, 3 Nov 1998, Mike Hackbart wrote:
>
>> problems with Novel network drives.  I have similar problems.  The problem
>> with not being able to create a new archive on a network drive was resolved
>> by plugging code from libc.a for the remove.c module into rcsedit.c and
>> doing a few little hacks.  I believe that I can reproduce this effort and
>> solve this problem too.  However something was done with printf and
>> changing EACCES to ENOENT that didn't seem quite right.  Were these issues
>> ever resolved?
>
>I believe they were.  The correct solution should be to let DJGPP's
>`remove' (or was it `rename'?) to do its job instead of using the
>work-arounds in the RCS sources.  (The RCS work-arounds are for other
>MSDOS compilers which, in DOS's tradition, won't delete/rename files which
>are write-protected.  The DJGPP's version doesn't have this problem, it
>works like the Posix standard requires, i.e. like Unix libraries do.)
>
>I don't remember the details, but there should be an option in the way 
>RCS is configured when you build it to tell it to simply call `remove'.
>
>If after that you still have problems, please post the details.  I really 
>cannot remember all the details after all this time ;-).
>
>And btw, make sure you use the latest version of the patched libc from 
>Nate Eldredge's site, since several functions in the stock library had 
>problems with Novell-mounted drives.

FYI, I have merely downloaded the binaries for rcs, in the hope that it
would work right out of the box.  It seems I have some work to do though,
since it will take some time for me to climb the djgpp learning curve and
figure out how to build rcs!  What is the URL for Nate's site, BTW?
>
>> the file checks out OK, it says locked and done, but, observing the
>> i:\soft\rcs\foo.c,v file using the Win95 Explorer, it changes to a folder
>> symbol and has size 0 bytes.  Now  an attempt to ci foo.c results in the
>> message:
>> ci: i:/soft/rcs/foo.c: Permission denied (EACCES)
>
>Forget how Explorer shows it, and tell what does "DIR" print about it, 
>both before and after you check-out the workfile.
>
>Also, please tell what command(s) were used to copy the file to the 
>networked drive.
>
To copy I used (originally I used Win95 drag and drop to do the copy, but
as it turns out it makes little difference.  However the name of the file
MAIN~1.C_V as given by the dir command seems a bit unusual):

c:\devel\rcstest\rcs>copy main~1.c_v i:\censoft\xxrcs

c:\devel\rcstest was then changed to delete the rcs folder and add a file
named rcs that contained:

i:\censoft\xxrcs

Then a dir in i:\censoft\xxrcs showed:

 Volume in drive I is SYS3       
 Directory of I:\CENSOFT\xxrcs

.              <DIR>                        .
..             <DIR>                        ..
MAIN     CV         34,814  11-03-98 10:33a main.c,v
         1 file(s)         34,814 bytes
         2 dir(s)   1,001,979,904 bytes free

Then in c:\devel\rcstest I did:

co -l main.c

The explorer picture did the same thing as before (changing main.c,v to a
folder), and another dir command in i:\censoft\xxrcs showed:

 Volume in drive I is SYS3       
 Directory of I:\CENSOFT\xxrcs

.              <DIR>                        .
..             <DIR>                        ..
MAIN     CV    <DIR>        11-03-98 10:31a main.c,v
         0 file(s)              0 bytes
         3 dir(s)   1,001,914,368 bytes free

and the resulting behavior with a new ci command at this point is the same.
 Now, having done this exercise, I realize that the times don't make sense.
 This has been a problem for us before and is due to not having
synchronization between the server time and the local PC time.  Could this
be the source of the problem?

Thanks for your help,

Mike

- Raw text -


  webmaster     delorie software   privacy  
  Copyright © 2019   by DJ Delorie     Updated Jul 2019