File Coverage

=head1 NAME

Autom4te::Channels - support functions for error and warning management

=head1 SYNOPSIS

  use Autom4te::Channels;

  # Register a channel to output warnings about unused variables.
  register_channel 'unused', type => 'warning';

  # Register a channel for system errors.
  register_channel 'system', type => 'error', exit_code => 4;

  # Output a message on channel 'unused'.
  msg 'unused', "$file:$line", "unused variable `$var'";

  # Make the 'unused' channel silent.
  setup_channel 'unused', silent => 1;

  # Turn on all channels of type 'warning'.
  setup_channel_type 'warning', silent => 0;

  # Redirect all channels to push messages on a Thread::Queue using
  # the specified serialization key.
  setup_channel_queue $queue, $key;

  # Output a message pending in a Thread::Queue.
  pop_channel_queue $queue;

  # Treat all warnings as errors.
  $warnings_are_errors = 1;

  # Exit with the greatest exit code encountered so far.
  exit $exit_code;

=head1 DESCRIPTION

This perl module provides support functions for handling diagnostic
channels in programs.  Channels can be registered to convey fatal,
error, warning, or debug messages.  Each channel has various options
(e.g. is the channel silent, should duplicate messages be removed,
etc.) that can also be overridden on a per-message basis.

=cut

=head2 Global Variables

=over 4

=item C<$exit_code>

The greatest exit code seen so far. C<$exit_code> is updated from
the C<exit_code> options of C<fatal> and C<error> channels.

=cut

=item C<$warnings_are_errors>

Set this variable to 1 if warning messages should be treated as
errors (i.e. if they should update C<$exit_code>).

=cut

=head2 Constants

=over 4

=item C<UP_NONE>, C<UP_TEXT>, C<UP_LOC_TEXT>

Possible values for the C<uniq_part> options.  This selects the part
of the message that should be considered when filtering out duplicates.
If C<UP_LOC_TEXT> is used, the location and the explanation message
are used for filtering.  If C<UP_TEXT> is used, only the explanation
message is used (so the same message will be filtered out if it appears
at different locations).  C<UP_NONE> means that duplicate messages
should be output.

=cut

=item C<US_LOCAL>, C<US_GLOBAL>

Possible values for the C<uniq_scope> options.
Use C<US_GLOBAL> for error messages that should be printed only
once during the execution of the program, C<US_LOCAL> for message that
should be printed only once per file.  (Actually, C<Channels> does not
do this now when files are changed, it relies on you calling
C<reset_local_duplicates> when this happens.)

=cut

=head2 Options

Channels accept the options described below.  These options can be
passed as a hash to the C<register_channel>, C<setup_channel>, and C<msg>
functions.  The possible keys, with their default value are:

=over

=item C<type =E<gt> 'warning'>

The type of the channel.  One of C<'debug'>, C<'warning'>, C<'error'>, or
C<'fatal'>.  Fatal messages abort the program when they are output.
Error messages update the exit status.  Debug and warning messages are
harmless, except that warnings are treated as errors if
C<$warnings_are_errors> is set.

=item C<exit_code =E<gt> 1>

The value to update C<$exit_code> with when a fatal or error message
is emitted.  C<$exit_code> is also updated for warnings output
when C<$warnings_are_errors> is set.

=item C<file =E<gt> \*STDERR>

The file where the error should be output.

=item C<silent =E<gt> 0>

Whether the channel should be silent.  Use this do disable a
category of warning, for instance.

=item C<ordered =E<gt> 1>

Whether, with multi-threaded execution, the message should be queued
for ordered output.

=item C<uniq_part =E<gt> UP_LOC_TEXT>

The part of the message subject to duplicate filtering.  See the
documentation for the C<UP_NONE>, C<UP_TEXT>, and C<UP_LOC_TEXT>
constants above.

C<uniq_part> can also be set to an arbitrary string that will be used
instead of the message when considering duplicates.

=item C<uniq_scope =E<gt> US_LOCAL>

The scope of duplicate filtering.  See the documentation for the
C<US_LOCAL>, and C<US_GLOBAL> constants above.

=item C<header =E<gt> ''>

A string to prepend to each message emitted through this channel.
With partial messages, only the first part will have C<header>
prepended.

=item C<footer =E<gt> ''>

A string to append to each message emitted through this channel.
With partial messages, only the final part will have C<footer>
appended.

=item C<backtrace =E<gt> 0>

Die with a stack backtrace after displaying the message.

=item C<partial =E<gt> 0>

When set, indicates a partial message that should
be output along with the next message with C<partial> unset.
Several partial messages can be stacked this way.

Duplicate filtering will apply to the I<global> message resulting from
all I<partial> messages, using the options from the last (non-partial)
message.  Linking associated messages is the main reason to use this
option.

For instance the following messages

  msg 'channel', 'foo:2', 'redefinition of A ...';
  msg 'channel', 'foo:1', '... A previously defined here';
  msg 'channel', 'foo:3', 'redefinition of A ...';
  msg 'channel', 'foo:1', '... A previously defined here';

will result in

 foo:2: redefinition of A ...
 foo:1: ... A previously defined here
 foo:3: redefinition of A ...

where the duplicate "I<... A previously defined here>" has been
filtered out.

Linking these messages using C<partial> as follows will prevent the
fourth message to disappear.

  msg 'channel', 'foo:2', 'redefinition of A ...', partial => 1;
  msg 'channel', 'foo:1', '... A previously defined here';
  msg 'channel', 'foo:3', 'redefinition of A ...', partial => 1;
  msg 'channel', 'foo:1', '... A previously defined here';

Note that because the stack of C<partial> messages is printed with the
first non-C<partial> message, most options of C<partial> messages will
be ignored.

=back

=cut

=head2 Functions

=over 4

=item C<reset_local_duplicates ()>

Reset local duplicate messages (see C<US_LOCAL>), and
return the number of messages that have been filtered out.

=cut

=item C<reset_global_duplicates ()>

Reset local duplicate messages (see C<US_GLOBAL>), and
return the number of messages that have been filtered out.

=cut

=item C<register_channel ($name, [%options])>

Declare channel C<$name>, and override the default options
with those listed in C<%options>.

=cut

=item C<exists_channel ($name)>

Returns true iff channel C<$name> has been registered.

=cut

=item C<channel_type ($name)>

Returns the type of channel C<$name> if it has been registered.
Returns the empty string otherwise.

=cut

=item C<msg ($channel, $location, $message, [%options])>

Emit a message on C<$channel>, overriding some options of the channel with
those specified in C<%options>.  Obviously C<$channel> must have been
registered with C<register_channel>.

C<$message> is the text of the message, and C<$location> is a location
associated to the message.

For instance to complain about some unused variable C<mumble>
declared at line 10 in F<foo.c>, one could do:

  msg 'unused', 'foo.c:10', "unused variable `mumble'";

If channel C<unused> is not silent (and if this message is not a duplicate),
the following would be output:

  foo.c:10: unused variable `mumble'

C<$location> can also be an instance of C<Autom4te::Location>.  In this
case, the stack of contexts will be displayed in addition.

If C<$message> contains newline characters, C<$location> is prepended
to each line.  For instance,

  msg 'error', 'somewhere', "1st line\n2nd line";

becomes

  somewhere: 1st line
  somewhere: 2nd line

If C<$location> is an empty string, it is replaced by the name of the
program.  Actually, if you don't use C<%options>, you can even
elide the empty C<$location>.  Thus

  msg 'fatal', '', 'fatal error';
  msg 'fatal', 'fatal error';

both print

  progname: fatal error

=cut

=item C<setup_channel ($channel, %options)>

Override the options of C<$channel> with those specified by C<%options>.

=cut

=item C<setup_channel_type ($type, %options)>

Override the options of any channel of type C<$type>
with those specified by C<%options>.

=cut

=item C<dup_channel_setup ()>, C<drop_channel_setup ()>

Sometimes it is necessary to make temporary modifications to channels.
For instance one may want to disable a warning while processing a
particular file, and then restore the initial setup.  These two
functions make it easy: C<dup_channel_setup ()> saves a copy of the
current configuration for later restoration by
C<drop_channel_setup ()>.

You can think of this as a stack of configurations whose first entry
is the active one.  C<dup_channel_setup ()> duplicates the first
entry, while C<drop_channel_setup ()> just deletes it.

=cut

=item C<buffer_messages (@types)>, C<flush_messages ()>

By default, when C<msg> is called, messages are processed immediately.

Sometimes it is necessary to delay the output of messages.
For instance you might want to make diagnostics before
channels have been completely configured.

After C<buffer_messages(@types)> has been called, messages sent with
C<msg> to a channel whose type is listed in C<@types> will be stored in a
list for later processing.

This backlog of messages is processed when C<flush_messages> is
called, with the current channel options (not the options in effect,
at the time of C<msg>).  So for instance, if some channel was silenced
in the meantime, messages to this channel will not be printed.

C<flush_messages> cancels the effect of C<buffer_messages>.  Following
calls to C<msg> are processed immediately as usual.

=cut

=item C<setup_channel_queue ($queue, $key)>

Set the queue to fill for each channel that is ordered,
and the key to use for serialization.

=cut

=item C<pop_channel_queue ($queue)>

pop a message off the $queue; the key has already been popped.

=cut

=head1 SEE ALSO

L<Autom4te::Location>

=head1 HISTORY

Written by Alexandre Duret-Lutz E<lt>F<adl@gnu.org>E<gt>.

=cut