On Tue, Mar 25, 2003 at 01:56:04AM +1000, Peter B. West wrote: > > > Victor Mote wrote: > > > > > >My understanding is as follows: > >1. The "refresh" button on the http://forrestbot.cocoondev.org site will > >take the contents at cvs.apache.org/cvs/xml-fop as input, and generate the > >web site. That is what is displayed at > >http://forrestbot.cocoondev.org/site/xml-fop. > >2. The "publish" button at http://forrestbot.cocoondev.org seems to > >actually > >update the cvs.apache.org/cvs/xml-site/targets/fop directory, checking the > >output from "refresh" into that repository. This is somewhat contrary to > >the > >text that appears on http://forrestbot.cocoondev.org. > > This step seems to have some problems. I have just gone through the > cycle of building locally, checking the changes into HEAD, getting > forrestbot to refresh, checking the just-constucted site, and then > getting forrestbot to publish. I had problems on the live site that I > did not see locally or on the forrestbot driver pages at > forrestbot.cocoondev.org. Specifically, an old copy of my javascript > file was in service.
The script which pulls files from CVS and publishes them is only run every X hours, so the live site probably just hasn't caught up. I've verified that the codedisplay.js files displayed by forrestbot.cocoondev.org and those in xml-site/targets/fop CVS are identical. > I have noticed that redundant files do not seem to be cleaned up in the > process either - they just hang around. Yes, detecting when a file *isn't* generated is harder than detecting when one is, and the price of making a mistake is much greater, so I didn't attempt it. --Jeff --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, email: [EMAIL PROTECTED]
