If the first line of a commit message is longer than 50 characters,
truncate it before adding the resulting string to the subject line of a
notification. This makes sure the subject line won't get too long
(unless the commit author name is unusually long, which we don't check).
The Git User's Manual recommends keeping the first line of a commit
message shorter than that, anyway:
| Though not required, it's a good idea to begin the commit message with
| a single short (less than 50 character) line summarizing the change,
| followed by a blank line and then a more thorough description. Tools
| that turn commits into email, for example, use the first line on the
| Subject line and the rest of the commit in the body.
[ http://www.kernel.org/pub/software/scm/git/docs/user-manual.html ]
Do not only generate notifications on commits, but also if a branch head
or lightweight tag was created, removed, or modified. Notifications on
branch head updates are omitted if one or more commit notification have
been generated and the branch head now references a descendant of the
originally referenced commit (which should be the usual case).
Most notifications include an ASCII "table" with two columns. The
formatting of these columns is now handled by the new format_table()
subroutine, so that the alignment can easily be changed in the future.
The code which handles notifications regarding tags was unused, as only
objects listed by git-rev-list(1) are considered, and git-rev-list(1)
never spits out the sha1 of a tag object.
Import the (self-written) git-update-mirror script, which updates clones
of Git repositories and then calls git-notify (in just the same way as a
post-receive hook would be called by Git). The git-notify script is
imported from git://source.winehq.org/git/tools.git (commit: 03d66f34)
and generates notifications on repository changes. We'll use these
scripts for generating our commit e-mails.
lines so this new test fails on the current head.
Note: check_snmp v1.4.13 with multi-line strings return somewhat v3 output;
it's not exactly what the specs say but it doesn't appears to break them
either. The fix could eventually supports both v2 and v3 output formats.
Only multi-line string test for now (regression test), counter rollover
tests planed with my snmp_counters_new branch.
NB: 64bit counters are broken in NetSNMP::agent from NetSNMP version 5.4.1
and lower, but might come in handy one day
Time differences are now set to 0 in case of backward timejumps so there are no wrap-around problems any more.
The RTA calculation hopefully gets a more accurate value in these cases also.
The checks for invalid/unavailable hosts need a complete command line.
They failed on hosts where no community was specified. Now, an invalid community is used so
the tests can run on hosts w/o snmp host/community in NPTest.cache. Execution will fail anyway.
Added snmp_interface and snmp_ifxtable test parameters to check_ifoperstatus.t.
Also getTestParameter calls for both plugins were converted to use three arguments
instead of undocumented >3 arg behaviour which led to redundant data in NPTest.cache.
