Re: [sldev] Notes from a Mac Build

2009-11-04 Thread Aimee Trescothick
You'd be better working off the Snowglobe branch at the moment which is more up to date. This and other issues compiling the viewer on the Mac with GCC 4.2 and against the 10.5 or 10.6 SDKs have already been fixed. See http://jira.secondlife.com/browse/SNOW-210 for this one. As far as I'm aware

Re: [sldev] Notes from a Mac Build

2009-11-04 Thread Lillian Yiyuan
You can tell the nice people developing the gcc that they made a mistake enforcing type checking then. However, you've provided no evidence that your position is based on any kind of knowledge. On Wed, Nov 4, 2009 at 5:55 AM, Argent Stonecutter wrote: > On 2009-11-04, at 07:34, Lillian Yiyuan w

Re: [sldev] Notes from a Mac Build

2009-11-04 Thread Argent Stonecutter
On 2009-11-04, at 07:34, Lillian Yiyuan wrote: > As I understand what is going on, it is definitely a vulnerability to > have a pointer to a fixed string being handed out as if it is a > variable. You can get rid of it by the flag for no write strings, > which is one of the options. But I don't kno

Re: [sldev] Notes from a Mac Build

2009-11-04 Thread Lillian Yiyuan
As I understand what is going on, it is definitely a vulnerability to have a pointer to a fixed string being handed out as if it is a variable. You can get rid of it by the flag for no write strings, which is one of the options. But I don't know enough about the problem to say which of the choices

Re: [sldev] Notes from a Mac Build

2009-11-04 Thread Argent Stonecutter
On 2009-11-03, at 12:30, Lillian Yiyuan wrote: > That produces the same error. I suppose the question is whether the > variables should be const char to begin with, but I don't know enough > about them to know whether they are modified, and thus don't want to > change the definition. The error also

Re: [sldev] Notes from a Mac Build

2009-11-03 Thread Lillian Yiyuan
That produces the same error. I suppose the question is whether the variables should be const char to begin with, but I don't know enough about them to know whether they are modified, and thus don't want to change the definition. The error also shows up in the mac-updater. On Tue, Nov 3, 2009

Re: [sldev] Notes from a Mac Build

2009-11-03 Thread Argent Stonecutter
Unless those values are actually going to be modified, that's an unnecessary function call. Also, if that code is executed repeatedly, it's a memory leak. Try making the conversion explicit with fakeconv.negative_sign = (char *)"-";

[sldev] Notes from a Mac Build

2009-11-03 Thread Lillian Yiyuan
Against the frozen viewer code, there was an error in llresmgr.cpp with "deprecated conversion of string to char*" I changed two lines: - fakeconv.negative_sign = "-"; - fakeconv.mon_grouping = "\x03\x03\x00"; // commas every 3 digits