[GRASS-dev] Re: [GRASS-PSC] GRASS 7 planning
hamish_b at yahoo.com
Sat Feb 9 21:31:07 EST 2008
Maris Nartiss wrote:
> > I don't want to ruin Your optimistic day, but there are yet 44 open
> > bugs and 9 (!) submitted patches in old bug tracker
> > (http://wald.intevation.org/tracker/?group_id=21).
Well yes, and there are 384 open bugs in the old RT system.
(we have already closed a lot of GRASS 5 bugs there as "wontfix", so
that number mostly applies to GRASS 6.x)
But many of those long-term bugs are still alive because they are not
simple to fix.
The majority of new bugs (trac) are likely to be quickly resolved so
are low hanging fruit and it should be a manageable task to deal with
e.g. trac bug #39 (r.in.srtm + zip) should be solved in days.
Similar are SegFaults which are usually quick to fix once discovered.
Some tickets do require more debate about architectural issues (e.g.
trac #7 [auto VAR files]) and may hang around for a long time. I don't
like releasing without that addressed, but hey, that one's not really
that major if the current version is functional.
> I have added the bugtracker links to
> If there are patches, they need to be tested and applied.
ie *evaluated*, tested, and if they pass those hurdles then applied.
for example, looking at some of the 9 mentioned above:
(these first two are mine; I would have commited them directly if I
though they were ready)
#372 scritps for converting raster maps into GRASS 7 format, and back
--dev experiment; doesn't apply
#374 v.what.vect additional upload options
--not sure if it's needed (see TODO section on 6.3 release roadmap
MediaWiki page). Unwilling to commit bloat for the sake of it.
#526 Provide option to end GRASS session from gis.m
--discussed on the mailing list several times in the past, without
(AFAIK) resolution about the best strategy.
> > I'm NOT proposing to delay 6.3.0. Just pointing out, that current
> > bug tracking system does not reflect current code quality status.
Right. I was just trying to focus on the new low hanging fruit.
Today's 6.3svn is up to >500k lines of code*. It's unrealistic to think
we can fix all bugs or wait for that to happen before release. I think
that is where the milestone targets in trac really help sort out the
true blockers from the "known-issues".
Some dumb metric:
384 RT wish+bug + 44 GForge + 14 open trac defects = 442 issues / 525k
SLOC = less than one issue per thousand lines of code. Not so bad.
In general (not directed at you Maris) I take issue with ideas that the
number of open bugs reflects on the current code quality status.
To me it seems that a high number of documented "known-issues" is a
strength and closing semi-fixed bugs just to get the number down is a
mistake. (that's not an exscuse not to try!)
>From a user standpoint I know when I find a bug in something if I find
an existing bug report for the issue I am reassured about the quality
of software, and maybe even get a work-around. If not I wonder what
else is broken.
[*] Results of SLOCCount 2.26 on latest svn/trunk:
Totals grouped by language (dominant language first):
ansic: 401306 (76.11%)
tcl: 45171 (8.57%)
cpp: 40649 (7.71%)
sh: 19359 (3.67%)
python: 18323 (3.47%)
perl: 1389 (0.26%)
yacc: 561 (0.11%)
lex: 480 (0.09%)
pascal: 38 (0.01%)
objc: 7 (0.00%)
sed: 1 (0.00%)
Total Physical Source Lines of Code (SLOC) = 527,284
Development Effort Estimate, Person-Years (Person-Months) = 144.27
(Basic COCOMO model, Person-Months = 2.4 * (KSLOC**1.05))
Schedule Estimate, Years (Months) = 3.54
(Basic COCOMO model, Months = 2.5 * (person-months**0.38))
Estimated Average Number of Developers (Effort/Schedule) = 40.72
Total Estimated Cost to Develop = $
(average salary = $56,286/year, overhead = 2.40).
lots of blah blah blah little coding from me today,
Never miss a thing. Make Yahoo your home page.
More information about the grass-dev