Matt Thomas | 3 Jun 08:36 2003

PCTS run on NetBSD.

After fixing a bug in make which the Open Group POSIX Conformance
Test Suite illuminated and banging on it to be happy with NetBSD,
I've done a preliminary run on NetSD 1.6R.  Some of my configuration
needs to change to eliminate some Unsupported errors but even so the
results are interesting.

The expanded result is at
ftp://ftp.netsd.org/pub/incoming/matt/PCSTS.results

BTW, having localedef instead of mklocale would be really nice.

--

-- 
Matt Thomas               Internet:   matt <at> 3am-software.com
3am Software Foundry      WWW URL:    http://www.3am-software.com/bio/matt/
Cupertino, CA             Disclaimer: I avow all knowledge of this message

Ben Harris | 3 Jun 13:01 2003
Picon

Re: PCTS run on NetBSD.

In article <5.1.1.6.2.20030602232710.090b5070 <at> 3am-software.com> you write:
>After fixing a bug in make which the Open Group POSIX Conformance
>Test Suite illuminated and banging on it to be happy with NetBSD,
>I've done a preliminary run on NetSD 1.6R.  Some of my configuration
>needs to change to eliminate some Unsupported errors but even so the
>results are interesting.
>
>The expanded result is at
>ftp://ftp.netsd.org/pub/incoming/matt/PCSTS.results

That's actually <ftp://ftp.netbsd.org/pub/incoming/matt/PCTS.results>.

Hmm.  Looks like I missed some namespace violations.  I'll fix them when I
get a chance, if no-one's got there before me.

--

-- 
Ben Harris                                                   <bjh21 <at> netbsd.org>
Portmaster, NetBSD/acorn26           <URL:http://www.netbsd.org/Ports/acorn26/>

mouss | 4 Jun 12:22 2003
Picon

Re: trailing / in fstab breaks mount <mount point>

This is because getfsfile() ($ROOT/src/lib/libc/gen/fstab.h) uses "dumb" string
comparison:
     if (!strcmp(_fs_fstab.fs_file, name)))
so "/var" and "/var/" won't match.
The same problem exists in getfsspec().

One way to fix the problem is compare after removing trailing all slashes
from the variables: name, _fs_fstab.fs_file and _fs_fstab.fs_spec.
So instead of using strcmp(), we would use a new function, say pathcmp(),
that compares canonical versions of its arguments.

any opinions?

At 02:31 31/05/2003 +0200, Nino Dehne wrote:

>It turned out I had to 1) remove my trailing slash in fstab:
>
>    /dev/wd0e   /var/   ffs   rw,softdep   0   2
>                    ^
>    /dev/wd0f   /usr/   ffs   rw,softdep   0   2
>
>or add to /etc/rc.conf:
>
>    critical_filesystems_local="/var/"
>    critical_filesystems_remote="/usr/"
>
>The actual problem is not with the rc.d scripts mountcritlocal and 
>mountcritremote I think but rather with mount(8) itself not accepting
>
>    mount /var
(Continue reading)

mouss | 4 Jun 12:30 2003
Picon

mount remounts... is PR 444 back?

I got the following behaviour of mount:

# mount /foo
# mount /foo
# mount
...
/bar on /foo ..
/bar on /foo ..

That is, mount remounts an already mounted fs (same mount point and same fs).
Is this is related to union fs or is it just PR 444 coming back?

A related PR is the #5805.

mouss

der Mouse | 4 Jun 19:41 2003
Picon

Re: mount remounts... is PR 444 back?

> I got the following behaviour of mount:

> # mount /foo
> # mount /foo
> # mount
> ...
> /bar on /foo ..
> /bar on /foo ..

> That is, mount remounts an already mounted fs (same mount point and
> same fs).  Is this is related to union fs or is it just PR 444 coming
> back?

It's not 444; that is about mount -a.

> A related PR is the #5805.

Quite so.

I don't see what the problem is.  When you mount something, it hides
what was there (or overlays it if -o union); if "what was there"
happens to be a previous mount, even "the same" mount, that doesn't
affect anything.

The only reason you don't get this with ufs (&c) mounts is that you
can't mount the same device more than once.  But you can mount
/dev/sd1f on /mnt and then /dev/sd2a on /mnt if you like.

"Unix doesn't prevent you from doing stupid things because that would
also prevent you from doing clever things."  I'm not quite sure what
(Continue reading)

John Hawkinson | 5 Jun 06:15 2003
Picon

[atatat <at> netbsd.org: CVS commit: src/usr.bin/cal]

Are we going to go adding color support to all of the basic unix apps?
It seems like a slippery slope, and I really question whether it is
one we should start down.

--jhawk

----- Forwarded message from Andrew Brown <atatat <at> netbsd.org> -----

From: Andrew Brown <atatat <at> netbsd.org>
Subject: CVS commit: src/usr.bin/cal
To: source-changes <at> netbsd.org
Reply-To: atatat <at> netbsd.org
Message-Id: <20030605002121.C60E3B004 <at> cvs.netbsd.org>
Date: Thu,  5 Jun 2003 00:21:21 +0000 (UTC)

Module Name:	src
Committed By:	atatat
Date:		Thu Jun  5 00:21:21 UTC 2003

Modified Files:
	src/usr.bin/cal: Makefile cal.1 cal.c

Log Message:
Add -h to cal, which makes it highlight the current date, if it's
present in the displayed calender.  It uses libtermcap to discover the
proper sequences to turn on bold, or uses overstriking if output is
not to a terminal.  If you use two -h options with terminal output,
the date is presented in reverse video instead of bold.

Next we'll have to make the Gregorian gap vary with TZ settings, since
(Continue reading)

Andrew Brown | 5 Jun 06:23 2003
Picon

Re: [atatat <at> netbsd.org: CVS commit: src/usr.bin/cal]

>Are we going to go adding color support to all of the basic unix apps?

what color?  this is "bold".  or "reverse video".

>It seems like a slippery slope, and I really question whether it is
>one we should start down.

hmm...maybe we should downgrade all man pages, then?

--

-- 
|-----< "CODE WARRIOR" >-----|
codewarrior <at> daemon.org             * "ah!  i see you have the internet
twofsonet <at> graffiti.com (Andrew Brown)                that goes *ping*!"
werdna <at> squooshy.com       * "information is power -- share the wealth."

mouss | 5 Jun 11:35 2003
Picon

Re: mount remounts... is PR 444 back?

At 13:41 04/06/2003 -0400, der Mouse wrote:

>The only reason you don't get this with ufs (&c) mounts is that you
>can't mount the same device more than once.  But you can mount
>/dev/sd1f on /mnt and then /dev/sd2a on /mnt if you like.

I have to update my head...
was almost certain that you couldn't mount the same fs on the same mount
point multiple times. Now, I indeed don't see any problems with this.

>"Unix doesn't prevent you from doing stupid things because that would
>also prevent you from doing clever things."  I'm not quite sure what
>the "clever thing" could be in this case, but people generally can't
>see them ahead of time....

agreed.
(That said, what's true for such clever things is also true for bugs, 
complications,
and such... one can't see'em ahead of time:-)

Thanks

mouss

Johan Danielsson | 5 Jun 16:04 2003
Picon
Picon

Re: [atatat <at> netbsd.org: CVS commit: src/usr.bin/cal]


Was there a good reason for choosing -h? It sounds like help to me.

/Johan

> From: Andrew Brown <atatat <at> netbsd.org>
> Subject: CVS commit: src/usr.bin/cal
> To: source-changes <at> netbsd.org
> Date: Thu,  5 Jun 2003 00:21:21 +0000 (UTC)
> Reply-To: atatat <at> netbsd.org
> 
> 
> Module Name:	src
> Committed By:	atatat
> Date:		Thu Jun  5 00:21:21 UTC 2003
> 
> Modified Files:
> 	src/usr.bin/cal: Makefile cal.1 cal.c
> 
> Log Message:
> Add -h to cal, which makes it highlight the current date, if it's
> present in the displayed calender.  It uses libtermcap to discover the
> proper sequences to turn on bold, or uses overstriking if output is
> not to a terminal.  If you use two -h options with terminal output,
> the date is presented in reverse video instead of bold.
> 
> Next we'll have to make the Gregorian gap vary with TZ settings, since
> the current method (do it only for September 1752) is decidely
> Anglo-centric.  ;-P
> 
(Continue reading)

Andrew Brown | 5 Jun 17:00 2003
Picon

Re: [atatat <at> netbsd.org: CVS commit: src/usr.bin/cal]

>Was there a good reason for choosing -h? It sounds like help to me.

the assocation with h and highlight seemed rather intuitive to me.

--

-- 
|-----< "CODE WARRIOR" >-----|
codewarrior <at> daemon.org             * "ah!  i see you have the internet
twofsonet <at> graffiti.com (Andrew Brown)                that goes *ping*!"
werdna <at> squooshy.com       * "information is power -- share the wealth."


Gmane