|
Chapter 38 Rule-Set Testing with -bt |
|
38.7 Add Debugging for Detail
In rule-testing mode the
-d
rule-testing command
can be used to turn debugging output on and off.
Prior to V8.7
sendmail
the
-d
could be specified only
on the command line. Beginning with V8.7
sendmail
, the
-d
may also be specified in rule-testing mode. We illustrate the latter
technique here.
Debugging output can reveal in great detail how individual rules are being handled.
A debugging category and level of
21.12
,
for example, causes
sendmail
to print the LHS of each rule as
it is tried.
To illustrate, consider the following (highly simplified) configuration-file
rule set:
S0 R@ $#local $:$n handle <> form R$*<@$+>$* $#$M $@$R $:$1<@$2>$3 user@some.where R$+ $#local $:$1 local names
Normal output that is produced when a rule-set number and address are
entered at the
>
prompt looks like this:
>0 georgerewrite: ruleset 0 input: george rewrite: ruleset 0 returns: $# local $: george
But if we turn on debugging using the
-d
rule-testing command:
>-d21.12
the output that is produced when the same rule-set number and address are entered is more verbose than it was before:
>0 georgerewrite: ruleset 0 input: george ---trying rule: @ --- rule fails ---trying rule: $* < @ $+ > $* --- rule fails ---trying rule: $+ ---rule matches: $# local $: $1 rewritten as: $# local $: george rewrite: ruleset 0 returns: $# local $: george
Observe that the first rule in rule set 0 (the lone
@
)
does not match
george
in the workspace. Therefore that rule
fails
and is skipped. Then the more complicated
rule (
$*<@$+>$*
) is tried, and it too fails.
Finally, the
$+
operator in the
last rule matches
george
, and the workspace is rewritten.
Note that the extra output that is produced by
-d
can potentially
run to many lines.
To capture the
output for later examination, consider running
sendmail
in
rule-testing mode from within a
script
(1),
emacs
(1), or
similar session.
Higher levels of debugging are also available for examining rules and
rule sets. The level
-d21.15
shows
$
digit
operators
on the RHS being substituted with values from the LHS. The level
-d21.35
causes each comparison, token versus token, to
be printed.
To turn off the extra debugging output, just reuse the
-d
rule-testing
command and specify a level of zero:
>-d21.0
A
-d
with no category or level behaves the same as
the
-d
command-line switch (see
Section 37.1, "The Syntax of -d"
). It sets
a default of
0-99.1
.
38.7.1 A Trick
In debugging large configuration files, the output that is produced
by the
-d21.15
switch can become too huge to examine
conveniently. A good alternative (when modifying or adding
rules) is to temporarily insert a fake subroutine call before
and after individual rules to see what they do:
R$* $:$>TEST $1fake subroutine call Rlhs rhs
With the fake wrapper around the new rule (the name
TEST
is
arbitrary), ordinary rule testing
with
-bt
now shows how the address is rewritten by that rule:
rewrite: ruleset 3 input: ...
rewrite: ruleset TEST input: ...
rewrite: ruleset TEST returns: ...
new rule acted here
rewrite: ruleset TEST input: ...
rewrite: ruleset TEST returns: ...
rewrite: ruleset 3 returns: ...
>
If you use this technique, remember, of course, to remove the fake subroutine calls before putting that configuration file into use.
|
|
|
|
| 38.6 Process-Specified Addresses |
|
38.8 Batch Rule-Set Testing |