Re: Bugs being closed faster than filed: Re: Wishlist for woody+1
On Wed, 29 May 2002, Anthony Towns wrote:
> What're you claiming as the current values for the number of archived bugs
> and the number of open bugs? As far as I can see your graph indicates
> that both values are slightly above 120000. Which, uh, doesn't appear
> to make any sense at all.
Well, the way I calculate is based on the last modified time of .log files.
As we get into the realm of currency, the bugs in the archive grow. Ie, bugs
that have not been modified in the last 10 days, are counted as part of the
archived set, for 10 days ago.
I suppose I should just stop counting archived bugs, after I get to now - 28
days.
> According to the CGI indexes, there're 26,662 open bugs at the moment and
> 94,523 archived bugs, making for a total of 121,185 bugs in the database.
> (The remaining bugs that get us up to #148386 were lost before we started
> archiving bug logs permanently, and haven't been recovered to date)
This combined total, 121,185 is easy to see as the green line on my graph.
However, I believe that it shouldn't be shifted. I'll fix the algorithm
tonight.
(as for the lost bugs, Joey Hess has a script that we use to import old,
static data(the bts used to generate static pages). Every once in awhile
someone finds an old archive of the bug system, and he runs that script, and
converts the files into the internal format debbugs uses).
> Your graph would indicate that we passed 100,000 bugs in the archive in
> early March when we haven't done that yet.
I'll look closer. I do stand by the total bug count, except for it being
shifted.
> > * load all bugs, active and archived:
> > * as each bug is loaded, do a date -= date % 86400, to round all dates down to
> > the current day.
> > * Additionally, for the modtime of .log, add 28 days.
> > * sort this list twice:
> > * by last modified time of .log
> > * by the open date, as recorded in .status.
> > * step thru the .log list, by day.
> > * For each stopping point, find the same point in the open sort
> > * add the current pointers to a stats list. Also, add the delta from the last
> > round.
> > * save the last 366 items, and send thru gnuplot.
>
> You seem, then, to be graphing the Bug# of the most recently filed or
> closed bug at the given date, which isn't really particularly informative.
No, I calculate the count of bugs at each day, given these 2 sorts.
> Sure. I exclude fixed bugs, closed bugs and wishlist bugs. It seems
> a pretty unwarranted assumption to think that the treatment of these
> bugs would be so completely different as to wholly reverse a trend that
> appears in the remaining bugs, which are by far the majority.
To go and great historical graphs, for each severity, would be rather complex,
as it would require parsing the data out of the .log file. I wish that file
was an easier format to deal with.
Maybe, if someone were insanely bored, ...
ps: I do have a .log to .mbox converter, but don't have a script to go the
other way, which would sorta be required to fit in with the current
software.
--
To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: