From: <th...@fr...> - 2005-10-08 10:49:03
|
> There seem to have not been many of serious bugs and fixes in the past year or > so... > > I'm just wondering whether everything's important is really working or > nobody cares about nasm beyond their own small needs that seem to be > satisfied :) sorry to chime in this late, but actually i did run across a latent bug that manifests since 0.98.37 in all versions. the change to -O parsing exposed a problem with optimizing=3 (which earlier couldn't be set through the command line), it can produce what looks like a truncated symbol table at least in win32 COFF output (which results in a linker error later). interestingly, the same file with -O2 (optimizing=2 since .37) produces a phase error (i thought i'd get a jump out of range error at most). i ran across it by chance in a private project where i used to assemble with -O3 (and which ended up as optimizing=15), and the same code assembles into a broken .obj when optimizing=3 (as it is the case since .37). only one file in the project is affected, and it's not even the biggest or most complex one. i can't promise to debug it anytime soon, but people may want to be aware of this. |