>For one, with rsync, if the network goes down, and an artifact that a >developer wants that hasn't yet been used is requested, nexus is useless. >He's still stuck. With an rsync everything is available.
Only if you run rsync fairly often and kill us with bandwidth. And does every home based developer allocate 30gb to hold their own copy of central? I don't see how this is better than a RM >Second, being a command line person, I like being able to login to the server >and do something like "find . -name "*.pom" | xargs grep "somestring"" and >such to find various things. (I know, the repo managers have search things, >but gui's suck) Nexus stores the files in a file system, you can do exactly the same if you choose to not use the lucene index. >Third, httpd can run and serve static files on some very lightweight hardware >that cannot even begin to consider running java. As such, it's much faster >than Nexus or others. I run Nexus in a vm with ubuntu jeos with 128mb ram for the whole vm on an old machine. Does it get much lighter than that? Nexus uses ~64mb of ram...even our repository.sonatype.org instance that gets slammed by all the Maven builds at ci.sonatype.org. >Finally, this is the most important thing to me, each "mirrored" repository >can be kept on a unique URL. http://proxy/central, http://proxy/java.net, >http://proxy/apache-incubator, http://proxy/apache-snapshot, etc.... Thus, I don't get what you mean here. All the repo managers expose the repos via individual urls, grouping is recommended but optional.
