[reportlab-users] cannot install reportlab using easy_install

[Robin Becker]

Sorin Sbarnea wrote:

> I did create a patch: http://nusunt.eu/download/libart_and_gt1_update.patch

> 

> Changes:

> * updated libart to the latest release

> * added #include <string.h> to gt1-namecontext.c

> 

> --

> /sorin

> http://nusunt.eu

.......

Thanks for the patch, but I'm not sure I want to apply it right now. I do have 
the latest code, but I don't think I always understand it.

eg in art_affine.c

-  if ((int)floor(x + EPSILON / 2) < 1)
+  if ((int)floor ((x + EPSILON / 2) < 1))


I don't understand this change at all; the first is reasonable ie is floor(v) 
less than 1 where v=x + EPSILON / 2; the latter seems to be taking the floor of 
a boolean expression ie floor(v<1); assuming that boolean expressions map to 0 
or 1 then I think the effect is unchanged, but it looks sort of dodgy. This code 
is used in some sort of postscript output, so probably doesn't hurt us.

On the other hand, many of the changes related to int and other casts were 
probably added by me to eliminate warnings from the win32 compilers.

If this stuff is to be fixed probably I ought to try and use the changes from 
2.3.12 --> 2.3.20 to adjust our code rather than just blindly use your patch.

Apart from any difficulties you had with compilation etc. did you have any 
specific problem that required updating this library?
-- 
Robin Becker