> 1. portability warnings for 64-bit processors. most of these are > generated by the PUSH/ALIGN macros.Most of those are bogus. I wouldn't bother, if I was you; use a better compiler, instead.> 2. float/double loss of precision warnings. most can be removed > with explicit casts or appending the 'f' specifier to constants.Those that can be helped like that are bogus. Don't know about the rest. Using explicit floats can generate (much!) better code sometimes though, so it might be a good idea to do this change anyway. Segher
hi. i joined the list a week ago. i'm using speexlib as the codec for GameSpy's new voice SDK [for voice chat for video games]. everything works well. kudos. i'd like to make a contribution to the codebase. i am bothered by the spew of warnings ( about 300 ) from my compiler. i typically frown on warnings anywhere in my codebase. COMPILER: Visual C++ .NET 2003 version 7.1.3088 WARNING LEVEL: Level 3 (/W3) with 64-bit compatability warnings enabled (/Wp64) the warnings primarily fall into two categories. 1. portability warnings for 64-bit processors. most of these are generated by the PUSH/ALIGN macros. 2. float/double loss of precision warnings. most can be removed with explicit casts or appending the 'f' specifier to constants. i am willing to fix all these warnings, but don't want to make the merge bothersome. i can probably turn this around in a day, but i want to know how to submit the fixes to ensure that they are backwards compatible with other peoples compilers, etc. since new point release just came out, i figured now would be the right time. let me know the right process for this. regards, jon _____________________ Jon Heiner Network Programmer Full Spectrum Warrior Pandemic Studios
> 1. portability warnings for 64-bit processors. most of these are > generated by the PUSH/ALIGN macros.If you find an easy fix, please tell me. Regardless, note that the cast is safe even if it may not look like it (i.e. I'm casting a pointer to an int, but it doesn't matter if the int is too small).> 2. float/double loss of precision warnings. most can be removed > with explicit casts or appending the 'f' specifier to constants.What version are you using? In 1.1.x I cleared many of those, though of course, some still remain. A patch would be welcome.> i am willing to fix all these warnings, but don't want to make the merge > bothersome. i can probably turn this around in a day, but i want to know > how to submit the fixes to ensure that they are backwards compatible > with other peoples compilers, etc. since new point release just came > out, i figured now would be the right time. let me know the right > process for this.The best would be to submit a patch using the "diff -u" format (standard patch format). Thanks, Jean-Marc -- Jean-Marc Valin <Jean-Marc.Valin@USherbrooke.ca> Universit? de Sherbrooke