Currently Hudson allows a job to be tied to a specific node or to a specific
label. It would be very helpful to be able to tie a job to a set of labels.

For example, consider I have nodes that are a mix of 32 and 64-bit machines;
Linux, Windows, and MacOS; with different compilers or environments (java 1.4,
1.5, 1.6; gcc 3.4, 4.0, 4.1, 4.3; python 2.3, 2.4, 2.5, 2.6). What I would like
to do is set up a job that builds on any node with

, and
another job that builds with, say

. Right now, I need
to create special-case labels for each permutation – which leads to an
impossibly large set of labels. For all permutations of this relatively small
example, each node would have 31 labels, and the whole system would have
thousands of distinct labels! This is clearly too cumbersome for any system
administrator to keep up with.

One way to implement the GUI might be to have a new section for "Tie this job to
specific labels:" which would bring up a set of check boxes, one for each
available label (or alternatively, a multi-select list box). The job could then
run on any nodes that contain all of the selected labels.