Re: Complex CVS repository - need help...

[Eric Sommerlade]

hi geir,

may be this solution is works for you:

with every commit in one module, you place a dummy file in a specific
directory to this module.
you set up a cron job that checks this for this file regularly.
if it finds it,
  it checks out and copies the module to the desired place, e.g. to the
staging server directories.
  it deletes the dummy file.

i set this up for automatic javadoc generation, since rebuilding the all
javadoc on
commit of a single file would be too slow. we also use it for regression
testing.

hope that helps,

eric




Geir Johannessen wrote:
> 
> 
> Under the /servers/ tree we run web servers in order to test and develop the
> portals before they are copied with ftp to the production web servers.
> 
> Our CVS problem arises from the fact that the lib-Nextra files must be
> placed under /servers/xxx.com/cgi-bin/lib for each of the portals
> (substitute xxx with the names of the portals). I cannot find anything in
> the FAQs or help sections that says how to handle this situation. If we
> could have had /servers/lib and /servers/xxx.com/cgi-bin/lib point to this,
> then everything would be OK. But unfortunately the web servers must run in a
> chrooted environment where they cannot access anything higher up than
> /servers/xxx.com/. Also symlinking is not allowed for security reasons. So
> the question is whether our directory structure can be addressed by CVS or
> if there maybe is another way to achieve what we want?
> 


-- 
Eric Sommerlade
CURRY. SMILE @ WORK. 
  
Curry Communication GmbH 
Behringstr. 28a 
22765 Hamburg 

TEL +49.40.398 344 00
FAX  +49.40.398 344 09