Subject: Re: something wrong with again....
To: None <>
From: BOUWSMA Beery <>
List: current-users
Date: 04/12/2002 21:45:31
%s wrote on %.3s, %lld Sep 1993

> > While we're mentioning this, can I agitate for a few minor fixes
> > to the CVS repository to clean things up slightly?
> > There are about 20-some files with trivial problems, some of which

> Can you tell which mirror you use, and which files you see corrupted ?

All which offer `cvsup' access -- jp, no, uk, de, and de2.
I haven't bothered to try other means of access to test other
sites yet...  Also, I'm mirroring the CVS repository itself,
rather than checking out the latest version or something from
the mirror.

There was one actual corrupted file on two sites, which has
since been corrected.  The ``corrupted'' files aren't damaged
per se, but contain apparent minor problems that prevent them
from being handled as efficiently as they could be (for example,
the entire file needs to be transferred)...

Anyway, let me just toss out the list as recorded by my latest
`cvsup' sessions.  The following list of files fail a loose CVS
RCS checksum test, and need to be transferred in their entirety.
When I do a `diff' between the saved `bad' versions and the new
ones, the only difference (in most cases) appears to be a `lock' --
this may be a peculiarity of `CVSup' not taking this into account,
perhaps -- I'm adding the developer of cvsup to the cc: line to
note this.

basesrc/libexec/cron/INSTALL,v: Checksum mismatch -- will transfer entire file
Bad version saved in /mnt/cvsup/NetBSD.cvs/basesrc/libexec/cron/#cvs.cvsup-342.1

The list of files with these checksum mismatches is as follows:

Applying fixups for collection netbsd/cvs
 Fixup basesrc/libexec/cron/INSTALL,v -> Attic
 Fixup basesrc/usr.bin/rdist/docmd.c,v
 Fixup basesrc/usr.bin/vi/docs/spell.ok,v -> Attic
 Fixup basesrc/usr.bin/vi/docs/,v -> Attic
 Fixup basesrc/usr.bin/vi/docs/vi.0.txt,v -> Attic
 Fixup basesrc/usr.bin/yacc/yyfix.1,v -> Attic
 Fixup basesrc/usr.sbin/sendmail/cf/ostype/bsdi1.0.m4,v -> Attic
 Fixup basesrc/usr.sbin/sendmail/makemap/Makefile.dist,v -> Attic
 Fixup basesrc/usr.sbin/sendmail/src/Makefile.DGUX,v -> Attic
 Fixup basesrc/usr.sbin/sendmail/src/Makefile.Dynix,v -> Attic
 Fixup basesrc/usr.sbin/sendmail/src/Makefile.FreeBSD,v -> Attic
 Fixup basesrc/usr.sbin/sendmail/src/Makefile.Mach386,v -> Attic
 Fixup basesrc/usr.sbin/sendmail/src/Makefile.RISCos,v -> Attic
 Fixup basesrc/usr.sbin/sendmail/src/Makefile.SCO,v -> Attic
 Fixup basesrc/usr.sbin/sendmail/src/Makefile.SVR4,v -> Attic
 Fixup basesrc/usr.sbin/sendmail/src/Makefile.Titan,v -> Attic
 Fixup syssrc/sys/arch/amiga/dev/device.h,v -> Attic
Shutting down connection to server

Most of these are due to a diff like this:

--- /mnt/cvsup/NetBSD.cvs/basesrc/libexec/cron/#cvs.cvsup-342.18774     Fri Apr
12 10:18:10 2002
+++ /mnt/cvsup/NetBSD.cvs/basesrc/libexec/cron/Attic/INSTALL,v  Thu Nov  4 21:42
:50 1999
@@ -8,7 +8,7 @@
-locks; strict;
+locks  mycroft:; strict;
 comment        @# @;

But there are one or two files listed above with a somewhat
different diff, like a delta out of order or something.  If
these `mycroft' locks shouldn't be there, it'll be easy to
see the remaining files that fail the checksum test, otherwise
I'll diff all the above to find the other problem(s) upon

Those are probably trivial fixes, if they are fixes needed
at all.  The following files are not recognized as normal RCS
files, and so, `cvsup' falls back to the `rsync' algorithm.

   SetAttrs syssrc/sys/dev/qbus/dhureg.h,v
*  Rsync syssrc/sys/dev/qbus/dl.c,v
*  Rsync syssrc/sys/dev/qbus/dz_uba.c,v
*  Rsync syssrc/sys/dev/qbus/uba.c,v
*  Rsync syssrc/sys/dev/qbus/ubareg.h,v

A quick look at these files suggests they aren't recognized as
RCS files because of something like

|   locks; strict;
|   comment @ * @;
->  strict  ;
|   1.59

But I'm just guessing here.  The other files don't seem to have
this particular line.  In the area of these files, I also get a
server error from all of the mirrors I'm trying to pull off of:

Server warning: Cannot open "/home/cvsup/cvsup/prefixes/NetBSD-cvs/syssrc/sys/de
v/qbus/dz.c,v": No such file or directory

But I have a copy of this file that looks like
46546 Feb  6 22:51 (CET)

All the cvsup-offering mirrors I use give the above error for this
file, and no others that I see.

These files seem to appear in another directory in the repository
and rather than the normal `SetAttrs' message when comparing
checksums, I get the following for five files:

*  Replace syssrc/sys/dev/dec/qbus/dl.c,v -> Attic
   SetAttrs syssrc/sys/dev/dec/qbus/dlreg.h,v -> Attic
*  Replace syssrc/sys/dev/dec/qbus/dz.c,v -> Attic
*  Replace syssrc/sys/dev/dec/qbus/dz_uba.c,v -> Attic
   SetAttrs syssrc/sys/dev/dec/qbus/dzreg.h,v -> Attic
   SetAttrs syssrc/sys/dev/dec/qbus/dzvar.h,v -> Attic
   SetAttrs syssrc/sys/dev/dec/qbus/files.uba,v -> Attic
*  Replace syssrc/sys/dev/dec/qbus/uba.c,v -> Attic
*  Replace syssrc/sys/dev/dec/qbus/ubareg.h,v -> Attic

It appears that these, like the above, have the extra `strict  ;'
line in them that appears to be the cause of these message as
best I can guess.  I haven't looked really closely.

I also seem to get the following messages from certain mirrors,
but not others, as not all of them carry the `htdocs' collection:

Server warning: RCS file error in "/usr/local/etc/cvsup/prefixes/NetBSD.cvs/htdo
cs/contrib/donations.html,v": Invalid next delta 1.57 for delta 1.49
Server warning: RCS file error in "/usr/local/etc/cvsup/prefixes/NetBSD.cvs/htdo
cs/contrib/donations.list,v": Invalid next delta 1.54 for delta 1.46

A cvsup checkout of this file seems to happen fine.

> I'm using the french mirror daily and didn't encounter a single file corruption
> since the rsync problem has been fixed.

I don't think these are actual corruption issues, but more of a
perfectionist issue.  I'm not sure how important they are, as a
bunch of these files are irrelevant for -current, but I thought
I'd check to see if these messages are significant and if anyone
wants to do minor cleanup of the master repository (where I guess
these non-problems originate) in order to quiet my updates and
make other problems more obvious, and make it possible to handle
these few files a bit more efficiently.

Feel free to ignore me if these are real non-problems or not
worth it, I just thought I'd mention them because I see them.

My apologies if there's a better list for issues like this --
I'm only following the -current list.

Also, once again, don't write me directly (I'll check the list)
since I promise to be offline, Real Soon Now, this time for sure!

barry bouwsma