The `farm' target run `server', but after the server starts, also
builds clients (via `ssh') as specified in a configuration file.
A client can be a VirtualBox virtual machine, in which case
the client machine can be started and stopped automatically.
Most of the work is in `distro-build/drive-clients' (in the
"distro-build" package), and that's where the configuration-file
format and client-machine requirements are documented.
If multiple values were expected and not provided
when type-checking a vector expression, TR would throw
an internal error.
Thanks to Pierpaolo for the report!
The 1-second granularity of filesystem timestamps is not
good enough to deetct changes to the file --- especially
when `raco pkg' installs links and runs `raco setup' (in
the same Rcket process), in which case a newly installed link
might not be detected.
The longer-term repair is probably to add file-changed
events to Racket (based on `inotify' and similar interfaces
provided by OSes).
The `if' case of `get-marks' was missing subscripts, which
caused it not to match and defer to an `if' in `get-marks-core',
but `get-marks-core' is not supposed to have an `if' rule.
Fix the one in `get-marks' and removed it from `get-marks-core'.
A collection declares release notes with a `release-notes' field in
"info.rkt".
The "doc" directory is now populated exclusively by generated content,
instead of having a static "release-notes" directory (and a "keep-dirs.rktd"
file to record the static directories).
Keep track of the out<n>.sxref timestamps, so that `raco setup'
can detect when the database is out of sync (instead of assuming that
it's always in sync with out<n>.sxref files).
The "racket" directory contains a pared-back version of the
repository, roughly.
The "pkgs" directory everything else in the repository, but
organized into packages.
I think that this should work, note that I used (racket-core -99) in the
info file for no good reason -- it should probably be properly placed
somewhere.
