Harlan Stenn | 13 Jan 05:50 2001

Re: Will's ark-world script

I figure Will is probably still on vacation.

I'm trying to run Will's script to test the state of the reveal dirs and
it's failing at the "import ark.cmdline" line.

Anybody got any ideas on what I need to do to make progress on this?

Obviously, I'm not a python hacker yet...

H
> =================================================================
> 
> #! /usr/bin/env python
> # 
> _VERSION = '$Revision$'
> _VERSION = _VERSION[11:-2]
> __doc__  = """
> 
> Find *everything* that is supposed to be revealed on this host (?),
> then see how that compares with what is in the reveal dirs.
> 
> ???ToDo
> """
> 
> import os,sys
> 
> # --------------------------------------------------------
> if __name__ == '__main__':
>     # Process the command-line arguments; 
>     # it will fail if anything bad happens:
(Continue reading)

Jos Backus | 13 Jan 09:15 2001

Re: Will's ark-world script

On Fri, Jan 12, 2001 at 10:49:54PM -0600, Harlan Stenn wrote:
> I'm trying to run Will's script to test the state of the reveal dirs and
> it's failing at the "import ark.cmdline" line.
> 
> Anybody got any ideas on what I need to do to make progress on this?

Maybe you need to tweak your PYTHONPATH or ARK_PYTHONPATH somehow? It looks
like python needs to find arkbase-py/ark/cmdline.py so I suspect that both
these variables should contain your specific .../arkbase-py directory.

> Obviously, I'm not a python hacker yet...

Me neither...

Hth,
--

-- 
Jos Backus                 _/  _/_/_/        "Modularity is not a hack."
                          _/  _/   _/                -- D. J. Bernstein
                         _/  _/_/_/             
                    _/  _/  _/    _/
josb <at> cncdsl.com     _/_/   _/_/_/            use Std::Disclaimer;
Harlan Stenn | 13 Jan 23:23 2001

Re: Will's ark-world script

Thanks!

The problem was that while ARK_PYTHONPATH was set, its value was not
being "passed" to python's sys.path .

I did a little grepping and figured out what to do.

H
Will Partain | 17 Jan 21:13 2001
Picon
Picon

*PYTHONPATH (was: Re: Will's ark-world script)

(Sorry, folks -- I came back.)

Jos B wrote:

> Maybe you need to tweak your PYTHONPATH or ARK_PYTHONPATH somehow?

and Harlan S replied:

> The problem was that while ARK_PYTHONPATH was set, its value was not
> being "passed" to python's sys.path .

First: I typically go after these 'import' annoyances by
invoking an interactive python interp (i.e. type "python"),
and noodling around from there [example below].  I sometimes
end up consulting the 'import' documentation --
http://www.python.org/doc/current/ref/import.html

Second: the *PYTHONPATH stuff in ARK *may well* be an
infelicity... Python's got to find the ARK code *somehow*,
and the ARK code has to find its initial info *somehow*, and
I don't think there's a perfect way.  We can hash it out
here.

Recall that my predilection is *against* relying on magic
things having to be right in a user's environment.  If a
user runs `/our/bin/netscape', it should Just Work; it
mostly certainly shouldn't depend on an exact setting of
LD_LIBRARY_PATH or the phase of the moon (much the same
thing, IMHO).

(Continue reading)

Unknown | 17 Jan 21:13 2001

*PYTHONPATH (was: Re: Will's ark-world script)

(Sorry, folks -- I came back.)

Jos B wrote:

> Maybe you need to tweak your PYTHONPATH or ARK_PYTHONPATH somehow?

and Harlan S replied:

> The problem was that while ARK_PYTHONPATH was set, its value was not
> being "passed" to python's sys.path .

First: I typically go after these 'import' annoyances by
invoking an interactive python interp (i.e. type "python"),
and noodling around from there [example below].  I sometimes
end up consulting the 'import' documentation --
http://www.python.org/doc/current/ref/import.html

Second: the *PYTHONPATH stuff in ARK *may well* be an
infelicity... Python's got to find the ARK code *somehow*,
and the ARK code has to find its initial info *somehow*, and
I don't think there's a perfect way.  We can hash it out
here.

Recall that my predilection is *against* relying on magic
things having to be right in a user's environment.  If a
user runs `/our/bin/netscape', it should Just Work; it
mostly certainly shouldn't depend on an exact setting of
LD_LIBRARY_PATH or the phase of the moon (much the same
thing, IMHO).

(Continue reading)

Will Partain | 17 Jan 23:08 2001
Picon
Picon

FYI: Arusha Project as an Agre "site"?

[Phil, see note after I sign off...]

Folks, below is an item from Phil Agre's esteemable Red Rock
Eater News Service, which I recommend generally (he reads
more in a week than I do in a decade).  One of his
big subjects is how technology fits/makes-sense in
social/institutional contexts, which is (of course) relevant
to deeper success for the Arusha Project.

In the item quoted, he suggests a notion of a "site"
(starting in the para beginning "I want to look at the
problem from another angle") for which the Arusha Project
might qualify.  By sharing teams' info about "how we did
stuff" (in stylized XML files), we hope to create the
illusion, for the lonely solo admin, of being part of a big
site with clever, robust solutions and clever, robust
(virtual) fellow admins around to help/explain.

Food for thought/stuff to get those neurons whizzing...

Will

[Phil, the Arusha Project is an in-an-early-stage
open-source effort that is seeking a way to do
*collaborative* Unix system administration, and to develop a
framework [code] to help w/ same.  Web site:
http://ark.sourceforge.net/  Not seeking publicity.]

==================================================================

(Continue reading)

Harlan Stenn | 18 Jan 06:27 2001

Re: *PYTHONPATH (was: Re: Will's ark-world script)

> (Sorry, folks -- I came back.)

Why sorry?  I'm glad you're back and I hope you had a great time!

> If we *must* have info coming out of the environment (as
> currently with ARK_TEAM and a couple of others...), then I'd
> rather they were blatantly ARK-specific (and few).

How about the *only* ARK envariable is the path to a config file that
has all of the rest of the info in it?

H
Will Partain | 18 Jan 23:47 2001
Picon
Picon

Re: *PYTHONPATH (was: Re: Will's ark-world script)

Harlan writes:

> > (Sorry, folks -- I came back.)
> 
> Why sorry?  ...

Well, you see, I think Matt was really hoping to stage a
coup d' projet; I think he has a secret desire to rename the
Arusha Project to the Tunbridge Wells Project (which is
probably where he's from...), after which he's change his
name by deed poll to Disgusted, so he can sign himself
"Disgusted of the Tunbridge Wells Project".  He's weird that
way. [*]

> How about the *only* ARK envariable is the path to a
> config file that has all of the rest of the info in it?

Yep, I'd go for that.  Whether life can really be that
simple, I don't know -- e.g. I kinda like doing 
"ARK_DEBUG=15 ark package ...".  If you want to hack it, or
throw a couple of paragraph of sketch our way, feel free.

Will

[*] If this paragraph makes no sense to you, just ignore
it.  It requires some UK cultural accretions to grok.
Will Partain | 19 Jan 22:29 2001
Picon
Picon

doing "regenerative" packages (ark 1)

Nothing new here, but... I ended up noodling a little w/
what you might call "regenerative" packages -- I would
*KILL* for a good name... -- and was pleased that it
worked.

First: a r-ve package is one where you routinely change the
"source" and need to deploy/reveal it over and over again.
Automount maps, export lists, /etc/sudoers, /etc/hosts, and
so on.  Your systems change, you change the "system source"
that records things, and you need to re-splat all the files
that depend on those facts.

In my case, I was doing automount maps and /etc/exports
files.  (I happen to generate these from "dchunks" xml
files, but it doesn't matter.)

Here's how I want these things to work:

* I want to regenerate all these files every night, but
  *not* put them into production [reveal].  Simple reason:
  these are the kinds of files that, if messed up (across
  every host), leave you with a very dead/hobbling system.

* The sidai stuff has a "diffing deploy" -- it goes right up
  to "deploy" on every host and then diffs that against the
  previously-revealed version.

* I want to look at those diffs before anything dangerous
  happens.

(Continue reading)

Will Partain | 22 Jan 22:50 2001
Picon
Picon

fresh ftp snapshots (ark1) uploaded

Folks, lest we have over-stale FTP bits, I've put a fresh
set in ftp://ark.sourceforge.net/pub/ark; the self-same
stuff is in CVS, too.  All vintage ARK (ark1) stuff...

Other than Deeply Obscure Bug Fixes in the glasli1 stuff
(i.e. one of my sites), the affixed ChangeLog entries cover
the of-slight-interest changes.

Will

diff -ru1 ark-20001210/ARK/arkbase-py/ark/ChangeLog ark-20010122/ARK/arkbase-py/ark/ChangeLog
--- ark-20001210/ARK/arkbase-py/ark/ChangeLog	Sun Dec 10 17:20:24 2000
+++ ark-20010122/ARK/arkbase-py/ark/ChangeLog	Mon Jan 22 21:20:56 2001
 <at>  <at>  -1 +1,12  <at>  <at> 
+2001-01-22  Will Partain  <partain <at> dcs.gla.ac.uk>
+
+	* xmldoc.py (ArkXmlDocHandler.fatalError): Make it work for both
+	pre- and post-2.0 Pythons.
+
+2000-12-11  Matt Holgate  <matt <at> dcs.gla.ac.uk>
+
+	* host.py (_HOST_FILES): Added stuff for gateways, nameservers
+				 and domainname
+	* added TemplateParser.py
+
 2000-12-10  Will Partain  <partain <at> dcs.gla.ac.uk>

=== sidai/arkbase-py/ark/ChangeLog:

2001-01-19  Will Partain  <partain <at> dcs.gla.ac.uk>
(Continue reading)


Gmane