Hello, I am working with Jakarta Gump (http://jakarta.apache.org/gump) and for the last couple of months it's primary output has been "down". Talk on the list has been how to resolve this, and hence my message here.
The gump output in question (at http://cvs.apache.org/builds/gump/latest/) is generated (I am told) on Sam Ruby's local machine, and outputs automatically uploaded. These automatic uploads started failing and Sam has been unable to find time to resolve the problems. Further, Sam is less involved in Jakarta Gump at the moment. What we'd like to see if an Apache install of Gump that is accessible to all gump committers [for maintenance], and is controlled by Apache. Reading http://www.apache.org/dev/machines.html#moof, it was proposed that I ask you if moof would be a good candidate for such an installation. Each run, Gump downloads a copy of the latest code of many projects, compiles these against each other, and produces documentation output. It generates e-mails and an RSS feed to notify project owners of issues. Gump is both a build and a test tool, it act as an early warning system for divergences/issues and keeps code re-use fresh. [Note: Today Gump runs *do* include non-apache code, e.g. junit, since apache code leverages this.] Here is "wishlist" (in Linux speak, don't know anything about MacOS.) Please note, I'm not familiar with how apache does things like this, nor am I much of a sys admin, so this is my attempt to transfer requests. Accounts: - A gump group with gump committers as members. [I assume a single account it frowned on, but I don't know how to manage things like cron via a group. I'd really like nothing to be under one persons control only, but I guess apache admin's can fix cron as root if needs be.] - SSH access to moof. Resources: - /opt/gump [maybe /home/gump, it is a dynamic beastie] to install Gump [and tools, e.g Forrest]. [du -s . = 64636.] - /var/gump to checkout source code, and compile. [du -s . = 7229412.] - /var/www/gump for WWW outputs [ du -s . = 385052.] - Network: One time (or infrequent) CVS download, and then updates (the changed portion) of the CVS dir [du -s . = 2051832.] each run. - CPU: Pretty intensive compilations for a number of hours [hard to quantify sorry, I don't know moof.] Note: It might be nice to keep a bit of history on the WWW site (Sam had a nightly folder) but that is resources dependent. Environment: - Python 2.3.2 - Java (or ability to install Java [maybe more than one JDK]) - Ability to install XML Forrest. (http://xml.apache.org/forrest) Network Environment: - An SMTP server (which allows outbound relaying). [Gump sends mail as if from project owners.] - An IP/DNS alias 'gump.apache.org' (if that is ok for a non-top-level project, otherwise whatever..) - An HTTP virtual server listening on the above DNS alias. - [Futures] Ability to write to directories (over the network?) that are mirrored. [Gump produced repository output.] I do not know if any of this is feasible/possible, but I'd appreciate your feedback/thoughts. Thanks in advance for your consideration/assistance. regards Adam -- Have you Gump'ed your code today? http://jakarta.apache.org/gump --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
