[Mailman-Developers] CVS / Queuing problem

Ken Manheimer klm@python.org
Tue, 6 Oct 1998 15:26:27 -0400 (EDT)

On Tue, 6 Oct 1998, Scott wrote:

> On Tue, Oct 06, 1998 at 02:42:45PM -0400, Ken Manheimer wrote:
> | I would be surprised if there aren't some other simple sanity checks
> | that would make the job of finding and correcting setup flaws much
> | easier...
> suggestions?

I'm interested in suggestions, too.  The kinds of things that spring to
my mind are mostly concerning existing data and resource files:

 - verify that the location and permissions of the set*id wrapper
   excecutables are as expected
 - verify that the locations of the enwrapped scripts are as expected 
 - check properness of permissions of existing maillist dbs - enable writing 
 - likewise existing log files, if existing, enable writing

These are just off the top of my head.  I drew the line at going into
internal data structures (like, not checking version settings on
maillist objects - that's up to the versioning stuff,
MailList.CheckVersion() and versions.Update()), and at examining
external system files (ie, aliases, pipermail).

Not that i would want to hold off on starting such a thing for these
considerations - they can always be added, and i think they ought be.
The trick in starting even a simple version of mmcheck, that just
checks the directory permissions, is getting the autoconf stuff to set
it properly from a mmcheck.in.  Or maybe it can get the settings from
the mm_cfg module - but then it would depend on *finding* the mm_cfg
file, not sure how independent the mmcheck should be.  (I'm inclined to
have mmcheck depend on mm_cfg, and be a big error if it doesn't get it,
period.)  This all is thinking out loud, tho!