On Tuesday, Aug 12, 2003, at 08:46 Europe/Rome, James Strachan wrote:
On Monday, August 11, 2003, at 08:54 pm, Noel J. Bergman wrote:
The only disagreement is - should all this dynamically generated stuff
go into CVS or not. It seems silly to put it into CVS - since its
trivial to regenerate the site at any point using Maven. Its worse than
saying lets check every version of a jar into CVS.
One of the reasons for putting dynamically generated artifacts into CVS is
that in the event of a recovery situation, CVS is on one server, and the web
sites are on another, so recovery can be relatively quick. It is
unreasonable to assume that the infrastructure team has the tools, expertise
or time to use each different publishing engine.
But the developers on the projects can easily do this for them? It takes about 5 minutes to completely redeploy the website. Is that too long?
it might take 5 minutes for them to do, but it might take a lot more just for the infrastructure team just to find the right person and ping him.
you might not be aware of the fact that some of the people in the Apache Infrastructure team have volunteered to be paged (24/7!!) to keep the infrastructure operational.
In case of an exploit where the web system is wiped out, they want to be able to restore everything within minutes and without depending on other people.
So the thru-CVS publishing policy is about having a backup and an easy way for them to republish stuff.
Now, can this be done better? that's for sure. but nobody ever volunteered to setup a better 'fail over' infrastructure for publishing (even if ForrestBot could be consider an attempt to do such a thing)
-- Stefano.
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
