Sorry, but you lost me. How does urllib come into the picture? You configure the job to take file parameter and name it foo.tmp. Then your build script checks if file foo.tmp exists. If it does, rename it to foo and start testing it. If it doesn't, build foo and start testing it.
-- Sami Andrew Melo <andrew.m...@gmail.com> kirjoitti 21.2.2012 kello 7.36: > Hey Sami, > > On Mon, Feb 20, 2012 at 11:32 PM, Sami Tikka <sjti...@gmail.com> wrote: > You just have to change your build scripts to use the uploaded file if it > exists in the workspace and do something else if it doesn't. > > I would like to do that, but it doesn't seem to work. If I make the call with > urllib (I'm using python), I get a 500 error back (so the job never fires). > If I add a file to it, the job works fine. If I remove the file > parameterization, it works. It's just when I try to submit a job and don't > include the file that things blow up. > > Unfortunately (well, fortunately in a way), I'm doing my qualifying exam > tomorrow AM, so I can't dig out the backtrace, but one of the stack frames > was in buildWithParameters. I can send that tomorrow once I'm done if that > helps. > > Thanks! > Andrew > > > -- Sami > > Andrew Melo <andrew.m...@gmail.com> kirjoitti 20.2.2012 kello 0.09: > > > Hello, everyone- > > > > We currently use jenkins to test commits that are in our Git > > repository, but I'd like to add support for developers to test their > > current workspace with jenkins before the commit makes it up. I added > > some logic to the beginning of the build process to unpack a tarball > > passed in with the file parameterization, which works great. However, > > any existing processes that try to submit a job end up failing if they > > don't pass in a file with the http request. > > > > Is there a way to have jenkins ignore if a file isn't passed in > > (perhaps making a zero-length file)? Or is there a better way to go > > about this? (Our job configuration is super complex, so maintaining > > two jobs with identical options modulo the build parameterization > > would probably end up with them desynchronizing). > > > > Thanks, > > Andrew > > > > -- > -- > Andrew Melo >
