>>>>> On Tue, 29 Nov 2005 19:24:27 +0100, =?ISO-8859-1?Q?Attila_F=FCl=F6p?=
>>>>> <[EMAIL PROTECTED]> said:
Attila> Martin Simmons wrote:
>>>>>>> On Tue, 29 Nov 2005 16:03:27 +0100, =?ISO-8859-1?Q?Attila_F=FCl=F6p?=
<[EMAIL PROTECTED]> said:
>>
>>
Attila> Martin Simmons wrote:
>> >>>>>>> On Tue, 29 Nov 2005 15:09:48 +0100,
=?ISO-8859-1?Q?Attila_F=FCl=F6p?= <[EMAIL PROTECTED]> said:
>> >>
>> >>
Attila> Kern Sibbald wrote:
>> >>
>> >> >> On Monday 28 November 2005 19:38, Martin Simmons wrote:
>> >> >>
>> >> >>>>>>>> On Mon, 28 Nov 2005 17:06:00 +0100, Kern Sibbald <[EMAIL
PROTECTED]>
>> >> >>>>>>>> said:
>> >> >>>
>> >> >>>
Kern> On Monday 28 November 2005 15:55, Martin Simmons wrote:
>> >> >>> >> >>>>> On Fri, 25 Nov 2005 09:21:29 +0000, Greg Cope
>> >> >>> >> >>>>> <[EMAIL PROTECTED]> said:
>> >> >>>
Greg> I am trying to glob a load of oracle backup files. As per Stephan
Greg> Ebelt's post these will be under /uNN dirs that are often on
>> >> >>>
>> >> >>> >> different Greg> file systems.
>> >> >>>
Greg> What I am trying to do is;
>> >> >>>
Greg> regex = "^\/u[0-9]{2}\/(oraback|oraarch)"
>> >> >>>
Greg> But the following fail to match anything bar '/'
>> >> >>>
Greg> regex = "^\/u[0-9]{2}\/(oraback|oraarch)"
Greg> regex = "^\/u[0-9]{2}\/oraexp"
Greg> regex = "^\/u[0-9]{2}\/oraback"
Greg> regex = "^\/u[0-9]{2}.oraback"
>> >> >>>
Greg> As expected regex = "^\/u[0-9]{2}" matches the lot.
>> >> >>>
Greg> Can anyone shed any light on this or the regex code?
>> >> >>>
>> >> >>> >> The problem is that Bacula walks down the tree of directories
looking
>> >> >>> >> for matching include/exclude patterns. When it finds an excluded
>> >> >>> >> directory, it doesn't look any deeper in that directory. Now
/u02
>> >> >>> >> doesn't match any of the above, so it stops the walk before
>> >> >>> >> /u02/oraback is found.
>> >> >>> >>
>> >> >>> >> BTW, I don't think you need \ before / in a regex.
>> >> >>>
Kern> Recently, I added something similar to this example to the FileSet
>> >> >>> chapter of Kern> the manual. It would be interesting to know what
you
>> >> >>> think about the example Kern> Martin since I haven't actually "run"
it. Of
>> >> >>> course, the example doesn't Kern> deal with more complex issues
such as
>> >> >>> traversing subdirectories where parents Kern> may be excluded ...
>> >> >>>
>> >> >>> I found the new examples, but what the ^.?*$ is "^.?*$" supposed to
mean?
>> >>
>> >>
Attila> If we are talking about extended regexp "^.?*$" indeed is a sytax
error. There
Attila> must be something in front of the *. You could say "^.?.*$" but this
is the
Attila> same as "^.*$".
>> >>
>> >> >>
>> >> >> Match the whole line no matter what it contains. .* doesn't seem to
work on Linux ...
>> >>
>> >>
Attila> Mhh if it doesn't work you're using basic regular expressions. (And
it means
Attila> a line containing any number of dots) And then "^.?*$" works and
means the
Attila> same as "^*$".
>> >>
>> >> The regexp syntax (basic/extended) is chosen then regcomp() is called
and
>> >> AFAICS bacula uses extended. The .* regexp filters the files correctly
for me
>> >> on FreeBSD.
>> >> There is a slight problem though: all of the directories are included,
even
>> >> those containing no included files. This is a design issue, not a bug
in
>> >> regexp matching.
>> >>
>>
Attila> Sorry, for the misunderstanding. I was referring to the fact that
"^.?*$" does not
Attila> work on FreeBSD and works on Linux, whereas "^.*$" does not work on
Linux. This is a
Attila> clear indication for me that for some reasons bacula on FreeBSD uses
extended regexp
Attila> but on Linux basic ones are used. I wasn't referring to your include
problem though.
>>
>> OK, I see.
>>
>> I just checked on Linux (glibc-2.3.3-74) and "^.?*$" doesn't raise a parse
>> error in extended mode. Not sure what they think it means though!
Attila> Dunno either. According to re_format(7): "A piece is an atom possibly
followed by a
Attila> single `*', `+', `?', or bound." So .?* should violate this
statement. Looks like
Attila> a bug, maybe the '?' is silently discarded, that would explain why it
works.
Yes, that's very likely.
Attila> Can you reproduce "^.*$" not matching any line, as Kern mentioned?
No, that always matches for me. Kern said "does not seem to work" so this
might be some other problem like the one I mentioned about directories.
__Martin
-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems? Stop! Download the new AJAX search engine that makes
searching your log files as easy as surfing the web. DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=7637&alloc_id=16865&op=click
_______________________________________________
Bacula-users mailing list
Bacula-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/bacula-users
