home
***
CD-ROM
|
disk
|
FTP
|
other
***
search
/
Amiga Format CD 42
/
Amiga Format AFCD42 (Issue 126, Aug 1999).iso
/
-serious-
/
programming
/
other
/
wild
/
appunti
/
bugs.txt
< prev
next >
Wrap
Text File
|
1999-05-25
|
4KB
|
82 lines
*** BUGS, fixati, da fixare, possibili, probabili, che vengono in mente...
- FIXED: A double opening of a module CAUSES BIG TROUBLES !!!
The NODE in list is RE-Inserted in the list, and causes big
troubles. This is the DOUBLE-Closing bug in the wild.library!
(to fix, do a check in load/kill module, add a wild opencount,...)
Now IT's FIXED, with the wm_CNT! Warning with old modules !
- FIXED: KillerBug in DisplayPlanarScreen: first, it was writing in the
Draw engine data, not in the Display !! And then, it allocated
the TmpRas, the AreaInfo and more with AllocVecPooled and freed
it with FreeMem ! Bleah.
- FIXED: Now, i have a working draw pipeline, so i can concentrate on the
3D checking of the Refs system. I viewed it was all wrong. I'm
looking for the bugs in TDCoreSimplyFast.
I've fixed one, where were exchanged the child and the parent link.
Then, fixed more: now seems ok. But never say...
- TO FIX: The DOUBLE-Closing BUG REVENGE !!! Maybe some modules too old, to
fix from the first time ? Or more conceptual bugs like the precedent?
Mah.
- FIXED: Another (new) bug in the closing: the tables killing!
Was because i changed the input of KillTable: no more pointing
on the struct wt_Table, but on the wt_Data !! So, when killing,
was forgot to add wt_Data to a1.
Probably, that was the DOUBLE-Closing revenge... but is to check.
- TO RECODE: In TDCoreSimplyFast, there is a bad-working test, the diagonals
edges outcheck: it removed more edges than is should do. So, just
for now, i removed it. When there is time, recode ! It causes bad
errors, especially with very big faces!
- FIXED: In TDCoreSimplyFast, recoded the OutFace check: 3 Hidden-> OutFace
and 2 Out-> OutFace! Caused BAD Crashes ! (Crash04 is ALSO for this
bug.)
- FIXED: In TDCoreSimplyFast, bugs in the BSP-Inserting routine!? The
BspPointers have a bad offset and trash-up the structure!?
!! It's AllocVecPooled who Trashes everything !! Because of
a bad write, the Pool is broken, so this routine does bas things.
Thanks to my MungwallLike system (just added) i found the badwrite.
It's in the BSPEntries memory: more faces than they should are
written here: SOME ARE INSERTED TWICE !!!
So, it's a BSP-Insert BUG !!! FindOut it.
Fixed! Was a very Unlucky case: if 2 sectors child were attached
to the same parent's face, the BspSave of this face was broken:
after the first attach, all ok, after the second attach, was
copyed on bspsave also the first attach. So, one sector was
kept attached even after the breakjoins cycle.
- TO FIX: There are more bad writes, probably in the Draw modules (hope no
Display!) because my MungwallLike system signals some hits (in
the Display module,why ? go to check,please!) with all the Flat
series (Fluff & Candy), and all the same hit.
- FIXED: A closing bug in DisplayPeter&James: if no screen, a lot of bad
free were done.
- TO FIX: If all edges are z-out in distance (may be in other cases also)
bad errors, like edges with no TmpBuffer but not flagged as bad.
If you have time...
- TO FIX: Peter&James can be dangerous if used by more than 1 app, because
of the variables inside the code (chunky buffer pointers, mainly),
you must reload them on every frame!
- FIXED: In Peter&James was used my AllocBitmap routine, but was useless and
buggy (memory crash), so i used the STDBITMAP screen type. Better
and safer.
- TO FIX: Still some bugs when faces are Z=0 (i think), they are splatted around.
- TO DO: In Tix+ broker the textures edges loop is very sure but very
unoptimized. When have time, do the rewrite.
- TO FIX: WildPrefs.library TestPrefs hang: the problem is in the Wild's
FindWildApp, sometimes finds an infinite loop: a wapp node pointing
to itself !! (to find out why happens...);
- TO FIX: DrawCandy+, don't work with hires ! Also DisplayTryPeJam+.library !