On Feb 25, 2007, at 8:14 AM, Alexey Verkhovsky wrote: > It would be insufficient, because you are talking about accessing > SVN repository from production boxes. In places we are talking > about, there is often a big, paranoid firewall (or two, or three) > between them.
I can't entirely disagree with that sentiment. At a former employer the makefile spit out tarballs of the final product, and marked those tarballs (and their contents) with the date, building user, and CVS tags related to that source. A subset of users had write access to the (semi-publicly readable) directory these tarballs were distributed from, and end-users could type a specific command to figure out what release they were running. This let support figure out exactly what they were debugging with. The company was doing software distribution rather than web services, but there might be some lessons there. Capistrano's current model for deploying new code is: local machine talks to app server, server pulls from svn repo I could see use for local machine exports from svn repo, scp's to app server All the other moving parts (disable_web, restart, symlink, enable_web, etc.) would be more or less the same. I'd be more comfortable with this approach, but don't need it right now. Which is to say: if someone builds it I'll want to play with it. Failing that, I may try to whip something up before the heat death of the universe. -faisal --~--~---------~--~----~------------~-------~--~----~ To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/capistrano -~----------~----~----~----~------~----~------~--~---
