Re: ark-world.sample problem...

On 27/5/01 19:49, Will Partain wrote:

>reveal ?  I don't even see how that's possible.  Every
>instance of LNDIR at least in sidai/package is in
><src-linkfarm> or <host-linkfarm>. ??? Inquiring minds want
>to know!

Sorry. I gave you a bum steer. It was something else failing and I wasn't 
reading it right.

I was having all that rhlinux weirdness last night so I bit the bullet 
and deleted the whole state/builds/deploys world and tried rebuilding it. 
With a couple of exceptions, I managed to  bootstrap my infrastructure 
with:

   .../boot/python .../arkbase/arkcmd package reveal \
      --force --use-deps=max -v ALL

The stoppers were some weird moments when it seemed to skip stages and 
get confused. I'm not sure yet what happened as I was fixing a bunch of 
mistakes as I went along and I may have made a meal of the state/builds. 
[Aside: half the problems I have seem to be fixed by deleting the state 
and build directories and trying again - --redo doesn't seem to do a very 
good job of ditching state sometimes.]

I've spent much of the day today trying to tidy up my XML files and get 
to grips with ARK internals. I've created a new thingtype to play with 
('website's). I had a lot of trouble getting this going in the first 
place. I kept running 'ark install --redo ark-site-gen' but it wasn't 
picking up my site-spec table. As for above, I found the trick was to 
junk the state and build directories. It was the Makefile not being 
regenerated.

I'm playing with the code to evaluate '@...@' bits. I'm trying to make it 
more generic so I can use it with my new thingtype. More news if I manage 
to produce anything sensible.

:-j


-- 
jonathan hogg, one good idea ltd, 131 queen margaret dr., glasgow g20 8pd
http://www.onegoodidea.com/ tel:+44-(0)7976-614338 fax:+44-(0)7970-537451