On 2/15/2011 6:13 PM, Jim Browne wrote:
>
> Hello,
>
> I'm new to Bacula and wanted to give some feedback based on my
> experience with the documentation and code. A few days ago I posted a
> proposed Documentation patch to the users list [1], which was re-posted
> by another to the devel list [2]. Apologies for not noticing earlier
> that the documentation and devel lists exist.
>
> I'm happy to open a Mantis issue with that patch if there is interest.
>
> Some further comments:
>
> It should be emphasized that for week-of-month calculations the week
> starts on Monday and ends on Sunday. That behavior is suprising to
> Americans, who start their weeks on Sunday. There is a mention of it,
> but it is at the end of the paragraph about week-of-year. I think it
> would save some trouble if it was mentioned again, separately as a part
> of a discussion about week-of-month.
>
> I also found the week of month keywords confusing. I interpret "first
> Monday" to mean the first Monday of a month, regardless of the week in
> which it falls (first or second.) What Bacula means is the Monday in
> the first week of the month, if it exists. I think it would help to
> include a clarifying statement and would be happy to write it if there
> is interest.
>
> Finally, there is a small bug in lib/btime.c. The code comment states:
>
> /*
> * Return the week of the month, base 0 (wom)
> * given tm_mday and tm_wday. Value returned
> * can be from 0 to 4 => week1, ... week5
> */
>
> But, tm_wom can return 5. Take January 2011, for example:
>
> Mo Tu We Th Fr Sa Su
> 1 2
> 3 4 5 6 7 8 9
> 10 11 12 13 14 15 16
> 17 18 19 20 21 22 23
> 24 25 26 27 28 29 30
> 31
>
> tm_wom(31,1) will return 5, which does not match any week of month mask.
>
> Calculations are:
>
> fs = (31%7) - 1 = 3 - 1 = 2
> fs> 0 (2> 0)
> mday> fs (31> 7)
> wom = 1 + (31 - 2 - 1) / 7 = 1 + 28 / 7 = 1 + 4 = 5
>
> So, I was curious if the documentation should mention that a Monday or a
> Tuesday can fall in a sixth week which can not be scheduled using
> week-of-month.
>
> Again, I'm happy to write patches for each of these issues and open
> mantis bugs for them. I'm just trying to determine if there is interest
> and general agreement.
I think the first step is download the source and try generating the
docs. I tried this a while back and failed. There was some issue with
my tool chain, I'm not sure what.
If you're keen to do that, it's a great start. Perhaps you'll find out
what I was missing. ;)
We do need people willing to update the docs. I have many documentation
changes I'd like to make. At present, my focus is elsewhere, but if we
can get the process working, I'd like to get back to it this summer.
>
> PS: I also found a bug in the scheduler and opened
> http://bugs.bacula.org/view.php?id=1699 with a report and fix.
>
> [1] http://sourceforge.net/mailarchive/forum.php?thread_name=20110213210436.GL51504%40centipede.turtlesys.net&forum_name=bacula-users
> [2] http://sourceforge.net/mailarchive/forum.php?thread_name=4D59C631.7010006%40langille.org&forum_name=bacula-devel
--
Dan Langille - http://langille.org/
|