[PATCH]Reduce messages in setup.log

Robert Collins robert.collins@itdomain.com.au
Mon Jan 28 22:14:00 GMT 2002

----- Original Message -----
From: "Michael A Chase" <mchase@ix.netcom.com>
> > Essentially, yes thats why they are there: incomplete methods. When
> > review your patch I'll be looking at the function status.
> There's no sense spending your time on the patch.  I need to change it
> drastically or withdraw it.

I appreciate this. I'll take you up on it as well :].

> If I understand the intent:
> 1.  If a method only returns a constant 0, the message should go to
> setup.log.full to make it easier to tell where more work may be

Yes, LOG_TIMESTAMP (so that a casual inspection will have it visible).
Likewise for any constant response, with a couple of exceptions:
* IF the method will *never* get implemented (ie, do we ever need
writeable archives?) Then the message level should be LOG_BABBLE, and it
should say something like "Foo::Bar - I wasn't expected to be
implemented OR used!"

> 2.  If a method does more than just return a constant 0, it should not
> log(), or if it does, the message is only needed in setup.log.

Yes, LOG_BABBLE for routine non-user-visible importance. Noting that a
function is called is usually not needed in fact - thus the // in the
compress_gz:: call we discussed earlier.

> Possible candidates:
>    Performs actual work: compress_bz::peek(), compress_gz::peek(),
> compress_gz::~compress_gz(), io_stream_cygfile::peek(), and
> io_stream_file::peek().

yah, drop to BABBLE, or just remove the logging completely.

>    Returns NULL: archive::next_file_name(),
compress::next_file_name(), and
> io_stream::factory().

factory is important, archive::next_file_name and
compress::next_file_name should never get called (ie LOG_TIMESTAMP
these) -  indicates a non-properly overriden sub class.

>    Returns -1: compress_bz::seek() and compress_gz::seek().

Right. Well, compress_* and seeking don't mix well IMO, so I'm happy for
these to become BABBLERS.


More information about the Cygwin-patches mailing list