Re: Are cylindrical worlds not fully supported?

[dancebob at mindspring dot com (Bob Carragher)]

Hans Ronne <hronne@pp.sbbs.se> sez:

> It runs fine on LinuxPPC with the most recent SDL version (1.2.2). I
> haven't used SDL 1.1.8 for a while but sdlconq used to run fine back when I
> did, too. In any case, I suggest you upgrade to 1.2.2 since there has been
> a lot of bug fixes since 1.1.8. The SDL code is evolving rapidly.

That fixed the crashing problem.  Thanks!

> The only way you can run sdlconq right now is through the command line.
> There is no popup dialog yet. You also have to specify what game to run on
> the command line (example: -g lord-rings). If you failed to do that,
> sdlconq used to crash. I thought this bug had been fixed and that it would
> default to the standard game now, but I could be wrong.

No, that is not fixed yet.  It crashes in the same location --
attempting to update variable sscreen, even though that
variable has not been initialized yet.

> >Unfortunately, even if one skips the GUI (by supplying a game
> >option, such as "-v"), the program dies in get_grid_surface()
> >the very first time update_screen() is called.  (The "format"
> >field of SDL_Surface variable, tmpsurf, looks corrupted.)
> 
> Does it still crash if you use -g to pick a game?

It no longer crashes there.  The upgrade to SDL v1.2.2 seems to
have fixed that.  I also don't need to use -g to specify the
standard game.  It is sufficient to specify at least one game
option.  (I use -v, since that's the most useful.)  That brings
up the game, and all is mostly well (there are some minor display
problems) until I try to explore the cylindrical world in a
cylindrical way -- i.e. by trying to scroll around the world
continuously.  This fails in that the scrolling ceases when I
reach either the western or eastern "end of the world."  (Note:
I modified the standard game to make the world's circumference
60 (that is, equal to the world's width), in order to achieve a
cylindrical world by default.)

Also, there seems to be a problem with the western end and the
manner in which the pointer is redrawn -- sometimes the (black)
mask is left on the screen.  That is, the underlying world hex
is not redrawn, leaving patches of black as one pulls the pointer
away from the western end.

Note that both of these problems also exist with the tcl/tk
version.  Perhaps there is a more fundamental problem with the
code, at least in the X/Unix version, that is independent of
the drawing interface?

				Bob