FreeNAS Code

Fix a potential corner case where the version would print out the $REVISION variable in parentheses (this mostly applies to release builds); see nanobsd/freenas-common:save_build() for more details about what I mean.

Cache the version to avoid unnecessary 'I/O' (even though it's just a memory disk).

Clean up 'find alias for smtplib' script:

1. Unbreak multiple recipient specification on the command line.
2. Make it apparent that -t actually is for parsing recipients from a message
body.
3. Use the email module with some minimal parsing to avoid unnecessary breakage
down the line because of ad hoc header parsing method previously employed.
4. Make send_mail a bit more robust / atomic.
5. Don't override library modules (email).
6. Remove dupe modules.
7. Simplify code a bit more (it could be simplified even further, but this is
good enough for now).

This will fix mass emailing, which affects SMART reports at least if multiple
recipients are specified.

This addresses ticket # 960.

use _charset for MIMEText

use smtplib.connect() if smtp auth was not provided

accept to send plain mime message

Build system, branding, and image size enhancements.

Overall:

- Use variables instead of hardcoded and relative path based assumptions.
- Change branding from FreeNAS to a (largely) build driven variable name to simplify muxing between FreeNAS and TrueNAS.
- Rename /etc/version.freenas to /etc/version .
- Change the version info to <sw-name>-<version>-<revision>-<arch>, e.g.
FreeNAS-8.1-r7817-amd64 .

build/do_build.sh:

- Simplify arguments for forcing builds again to -f (no args), based on feedback from jpaetzel.
- Use variables as necessary.
- Add alpha support for unionfs based builds. This doesn't work with UFS SU+J on 9.0-BETA1+, so caveat emptor.

build/functions.sh:

- Consolidate build-related functions into a single file for simplicity.

build/nano_env:

- Add a file which contains the master info for how a build is performed.

gui/...:

- Chase the changes done above, but try to simplify / improve how things are done, emitting more diagnostic info than before.

nanobsd/common:

1. Allow the end user to specify the number of CPUS in the environment if they want, instead of hardcoding $CPUS + 1.
2. Nuke .pyc files. They're created automatically by They'll get regenerated on the fly once they're read in again by the interpreter.
3. Nuke the headers (minus the pyconfig.h -- it's required at runtime) since we're not compiling binaries (cc, etc isn't included in the final image).
4. Nuke all of perl the right way if $DEBUG == 1.

Misc:
nanobsd/freenas-common:

- Start features based build tunables for

syslog common.system.send_mail errors in /var/log/maillog

extend send_mail a bit with backward compatibility

- Simplify some logic.
- Improve style a little bit.