CF: Pupland and buggy bugfixes

[Christian Stieber <stieber@informatik.tu-muenchen.de>]

Mark Wedel (mark@icp.siemens.com) wrote:

> > In fact, pupland doesn't have many bugs at all :-) The big problems
> > have always been caused by quick fixes to some minor problems in the
> > server.

> Pupland made use of 'features' which are bugs on other maps.  In
> some sense it is just timing - if the fixes were made to the server
> ealier on, pupland would never have been able to use the features
> and no problem would be there.

Actually there is no way to know which features are features and which
are bugs since there is no documentation. So pupland does indeed make
use of features ("show invisible" is the best example --- how should
the pupland designers have known that objects are not intended to be
invisible? That's something you came up with after we reported the
magic mouth problem, to get a quick fix instead of doing it properly
(three solutions: a flag, putting magic mouths below the floor on
maps, or just changing the display code so that the password for magic
mouths is no longer displayed). A quick fix was hacked instead which
broke pupland, for several versions of crossfire. It's a great example
for bad software/game design, though, so the fix has a good side
as well).

Granted, this is the worst "bugfix" that crossfire has seen since
0.93.4 :-) There have been quite a few other "fixes" that made things
worse.

Reading about the 0.95.x problems caused by bugfixes (usually of problems
that nobody here has ever noticed) just makes me think "oh, again".
It's generally the same scheme: something is "fixed" in the server.
The fix is either buggy and never tested (like the singing-abuse-fix
(no experience at all), the polymorph fix (powerful items could be
created), the "only use keys from keyrings, not bags" fix (server
crashed when running into a door) and others: obvious problems, not
some weird thing), or the fix affects maps that are not changed (show
invisible or the pit-"fixes").

Of course there have been real bugfixes as well, but one tends to
remember the ones that made things worse (usually turning a minor
problem into a major problem).

Sorry for this... I'll try to shutup now.

Christian


-- 
Christian Stieber        http://www.informatik.tu-muenchen.de/~stieber
-
[you can put yourself on the announcement list only or unsubscribe altogether
by sending an email stating your wishes to crossfire-request@ifi.uio.no]