[crossfire] Modelling monsters, longer animations?

[Mark Wedel]

  To me, the model sizes isn't that big a deal.

  After all, most people will only need to check out the arch tree once at most. 
    So that initial checkout will take longer.  But unless the models actually 
change, the time to do cvs updates shouldn't be longer.

  The only way to get around that is for the models to be in a completely 
different repository (arch3d) which seems odd to me).

  As for generation, not required that makefiles be used, but the method to 
update them needs to be pretty clear.  If I have to run some command and pass it 
5 options for it to do the right thing, that is hardly appropriate.

  That said, properly written makefiles shouldn't have any of the problems 
tchize mentions - it should only create new images if the model itself has a 
later timestamp that the image it generates.  So there shouldn't be any problem 
with someone typing 'make models' and it generating all new images - it should 
only generate those out of date, which is what should happen.

  That said, CVS does have timing issues - if the check out order isn't write, 
the makefiles can thing the images are out of date.

  However, any update mechanism potentially has that problem - the model looks 
newer than the images.

  Slightly more complex makefile would do something like 'if out of date, 
generate as png.new, compare with old png, if different, mv png.new to png, else 
remove png.new'

  The only problem there is that the script will keep trying to regenerate the 
image because it will think it is out of date.  And if you do a 'touch 
old_image.png', cvs will think it is different and want to commit it (note that 
this isn't an issue related to binary vs text files - cvs commit doesn't look at 
contents, it just looks at the date stamp)