From: Richard R. <sf...@ol...> - 2004-09-06 07:38:17
|
On Thu, Sep 02, 2004 at 03:16:04PM -0700, Peter Montgomery wrote: [...] > > - Remove the "freeglut_assert_ready" and > > "freeglut_assert_window" invocations from places where the code can do > > something reasonable even if the assertion would fail [...] > I know there is always a desire to remove assertions from functions that > have been working for years or that have error handling code in them > already. However, the assertions cost nothing for the release build, and I think that this relates to the "assert abuse" comments that I attached to many assert() uses in OpenGLUT. I have personally not seen or felt the desire to remove an assert() just because "the code seems to work". I cannot speak for anyone else, of course. The rationale has nothing to do with a false sense of security (and it surely would be false). Rather it is simply that there are many places in freeglut where assert() is abused as if it were a general error-checking mechanism. More is explained on the openglut-devel list, where I can actually speak for a developer (myself). OpenGLUT inherited the freeglut uses of assert(), and I consider this an issue requiring solution. --=20 "I probably don't know what I'm talking about." http://www.olib.org/~rkr/ |