Re: [Pipmak-Users] Unofficial Release of Pipmak Build 155 for Win32

[Christian W. <cwa...@gm...>]

[It seems this message didn't make it through on the first attempt - 
let's try this agin.]

Andrea Viarengo wrote:
> I also write an how-to in the Pipmak Wiki
> 
>   http://pipmak.sourceforge.net/wiki/index.php/MSVC8_Build
> 
> and in this page
> 
>   http://pipmak.sourceforge.net/wiki/index.php/Source_Tweaks_on_SVN_Build155
> 
> there are all the changes that I have made to source code to pass
> compilation stage

Very cool, thanks!

I've tried to fix some of the warnings. Can you apply the attached patch
to revision 156 and tell me what warnings you still get? Particularly
the ones about float/double conversions (I haven't found a warning
setting for this for GCC).

I see you use Lua 5.1. Because there are a few changes in the language,
I haven't switched Pipmak from Lua 5.0 (latest release is 5.0.3) to 5.1
yet. I intend to do that with Pipmak 0.3 when I can break backwards
compatibility. I don't know if there have been any API changes that
could explain your problems, but you might want to try Lua 5.0.

> The main change to the source code is made on PipimakLuaLib.c 
> I have had to commented out the calling to the function:
>   checkinstanceof(L, 1, "pipmak-node", "node");
> I haven't notice any controindication caused by the absence of this calling, 
> perhaps only Christian know ...

This is just error checking. Without that call, you could call a node
method on something that's not a node, with probably unpredictable
results (including crashes). It checks if the object at index 1 in the
Lua stack has a metatable called "pipmak-node", or has a metatable whose
metatable is called "pipmak-node", etc., and if not, stops with an error
message about wanting a "node". The names of metatables are stored in
the Lua registry table, with the metatables themselves as the keys.

> I notice some differences on patches locations 
> (this project still use old syntax)
> The patches are smaller of 1 or 2 pixel respect official pipmak 0.2.6 
> and I see white lines beetween patches.

That must be a bug. Patches specified with the old syntax are not
supposed to look any different than before. Can you send me a screenshot
of what you see in the new version and how it looks in the old version?
If you want to be sure it's not something about the different compilers
that causes this, you could also try updating to revision 154 (before
the arbitrary placement) and recompiling.


   -Christian