http://issues.apache.org/SpamAssassin/show_bug.cgi?id=4824





------- Additional Comments From [EMAIL PROTECTED]  2006-03-20 15:47 -------
Created an attachment (id=3423)
 --> (http://issues.apache.org/SpamAssassin/attachment.cgi?id=3423&action=view)
implementation

to recap - if you have

  - a plugin .pm file in your sandbox
  - a .cf file that calls "loadplugin YourPlugin.pm"
  - and rules that use the plugin
  - but the .pm file is not listed in the MANIFEST file

build/mkrules will ignore the plugin. Instead it gives this error:

log/mkrules_t/rulesrc/sandbox/foo/20_foo.cf: WARNING:
'log/mkrules_t/rules/plugin.pm' not listed in manifest file, line ignored:
loadplugin Good plugin.pm

That's OK for plugin files that are publishable and ready to go.  However,
it's bad for "sandbox plugins" -- ie. plugin code that you're hacking
on and expecting to be experimental.  What it should do is:

  - if the .pm is in the MANIFEST, all is good; copy to rules as normal

  - if the .pm is not in the MANIFEST, it should still allow the rules to run
    in development mode, but not once the code is 'distributed' (ie. no
    longer in development).

Those terms defined per http://wiki.apache.org/spamassassin/RuleLifeCycle:

development mode: builddir, make_test, mass_check, bbtest, nightly

distributed: make_install, tarball, sa_update, also "make disttest"

(Normally, the "bbmass" state also would be considered part of the
'development' set, but sandbox plugins are currently blocked in that state --
however this is done for other reasons.)

This patch implements this feature, by adding a new command --
"tryplugin".  It's the same as "loadplugin", but if the plugin file
cannot be read, this is silently ignored.   "build/mkrules" will
then convert "loadplugin" lines for plugins not listed in MANIFEST,
into "tryplugin".

Patch is for review -- the change is in trunk:

Sending        build/mkrules
Sending        lib/Mail/SpamAssassin/Conf.pm
Sending        lib/Mail/SpamAssassin/PluginHandler.pm
Sending        t/mkrules.t
Transmitting file data ....
Committed revision 387224.




------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

Reply via email to