[9fans] Funny dates

[9fans] Funny dates

[Sam]

I have some funny dates on files/directories
in my filesystem.  Some are directories we
may have touched, but many are those we know
we haven't (eg bitsy/*):

	term% cd /sys/src/9;ls -l
	d-rwxrwxr-x M 9 sys sys   0 Dec 12 13:43 alphapc
	d-rwxrwxr-x M 9 sys sys   0 Mar  6 12:59 bitsy
	d-rwxrwxr-x M 9 sys sys   0 Mar  8 06:36 boot
	d-rwxrwxr-x M 9 sys sys   0 Nov  8 12:03 ip
	--rw-rw-r-- M 9 sys sys 191 Apr  2  2002 mkfile
	d-rwxrwxr-x M 9 sys sys   0 Nov  8 11:55 mtx
	d-rwxrwxr-x M 9 sys sys   0 Mar  8 06:51 pc
	d-rwxrwxr-x M 9 sys sys   0 Mar  6 13:19 port

All of the erroneous dates seem to be in the Nov-Dec
time frame.  This causes a few irritating, erroneous locally
modified messages on pull.  Before I castigate replica
for not playing nicely in my filesystem can anyone
suggest what I can do to correct this?

Cheers,

Sam

[9fans] Replica (was: Funny dates)

[Sam]

Ahem.

My apologies - clearly I didn't understand
ls's
	/* 6 months in the past or a day in the future */
which I now do.  Improper error attributions
to replica revoked; they're likely my fault.

Two replica notes on source:
Suppose I mk something and don't bother
removing the .8s.  I do a pull and get
a new source file.  The .8
can be newer than the source and the
next mk won't recompile the changes.
This suggests a mk clean is necessary
before each mk.

Apparently one should never do a mk
install unless you really want to
replace the existing functionality
with a local implementation.  If you
accidently do, the binary is created
as newer than sources and until you
force pull the binary the source gets
updated, but the binary doesn't.

I've been thinking about a way to tinker
with the source without getting mired
in force pulling source files.  Perhaps
this is obvious, but I'm going to start
making source changes by:
	mkdir local
	cp x.c local
	(make changes to local/x.c)
	bind -c . .
	bind -b local .
	mk
this way a simple shell script will let
me walk a tree binding local directories
as above; then I can build, eg, local
kernels without breaking sources consistency.

Is this similar in nature to what others do?

Cheers,

Sam

On Sat, 8 Mar 2003, Sam wrote:

> I have some funny dates on files/directories
> in my filesystem.  Some are directories we
> may have touched, but many are those we know
> we haven't (eg bitsy/*):
>
> 	term% cd /sys/src/9;ls -l
> 	d-rwxrwxr-x M 9 sys sys   0 Dec 12 13:43 alphapc
> 	d-rwxrwxr-x M 9 sys sys   0 Mar  6 12:59 bitsy
> 	d-rwxrwxr-x M 9 sys sys   0 Mar  8 06:36 boot
> 	d-rwxrwxr-x M 9 sys sys   0 Nov  8 12:03 ip
> 	--rw-rw-r-- M 9 sys sys 191 Apr  2  2002 mkfile
> 	d-rwxrwxr-x M 9 sys sys   0 Nov  8 11:55 mtx
> 	d-rwxrwxr-x M 9 sys sys   0 Mar  8 06:51 pc
> 	d-rwxrwxr-x M 9 sys sys   0 Mar  6 13:19 port
>
> All of the erroneous dates seem to be in the Nov-Dec
> time frame.  This causes a few irritating, erroneous locally
> modified messages on pull.  Before I castigate replica
> for not playing nicely in my filesystem can anyone
> suggest what I can do to correct this?
>
> Cheers,
>
> Sam
>
>