Nicola Ken Barozzi wrote: > > One thing I like about Forrest is that it's not only a static site > generation engine, but it's capable of serving the site live. Changes > are instantaneous, bandwith is perserved, and dynamic content can be > employed. > > IMHO every Apache project that produces something it can use for itself > *should* actually use it, like HTTPD does. This gives the team something > to work collectively on, and a base that keeps the project adherent to > actual needs. > > IOW, I would like to see the Forrest website served of a live Forrest > instance. > > To do this, even before talking about logistics and permissions, we have > to be sure that it's not going to cause problems, even in high load, and > be able to administer it. > Before we can seriously start to consider using it live, we should at > least have the following: > > 1 - knowledge of the memory it's going to need and use > 2 - have no known memory leakage (to be tested) > 3 - have caching turned on > 4 - know what to do when upgrading the version > 5 - decide where it's going to get the site from > 6 - other? > > After that, it would be beneficial to be able to serve multiple sites on > a single Forrest instance, so that we can have other projects join and > have their site served. > > Thoughts, comments, opinions?
This is a RT thread so let us all throw our Random Thoughts into the mix. We can later develop that into a plan. I see various different issues. Some are separate, and some are entwined ... building our website (static, dynamic, both); a place to experiment with our own software; a testing environment; a way to stage the website so that we can review our changes before publishing; running a Forrestbot to build static websites for various ASF projects; etc. We will have a place to experiment soon. Each PMC will get $pmc.zones.apache.org ... they are still testing the setup, so there is no rush. That would be our space to experiment with our various versions of Forrest. We could have the release version, the head of trunk, and a "known good trunk". I am not talking about versions of the docs, this is about versions of the software. We will need to experiment there, until we are happy to use it in production. We don't don't need to worry too much about memory usage at this stage. In fact, that zone will give us a place to do proper tests to determine the memory aspects. For our actual website, there are many things that we need to consider. For example, ASF Infrastructure has certain requirements to enable easy recovery of websites and a way to fail-over to another server to enable OS upgrades etc. I think that the main thing is to do it in steps. Start with the demo servers in forrest.zones.apache.org and work up to doing more ambitious things like the website. There is another related issue. Over the past years, a plan has come together to have a "staging" server, i.e. a place where static sites are built and previewed before going into production. That is now a lot closer to happening. See the threads at [EMAIL PROTECTED] about "Site Build Zone", and also http://forrest.apache.org/0.7/proposal-asf-publish.html Of course, the Forrestbot is a part of that infrastructure and they want Forrest committers to be involved in that. Anyway, the zone is not ready yet, so i will keep us informed. --David