Allright, although unfortunately I will need to have TSReleaseFatal() and TSFatal() to correspond to TSReleaseAssert() and TSAssert() (both of which are used in this style). I'll write up a man page to send around.
On Fri, Feb 16, 2018 at 3:30 PM, Rushford, John <john_rushf...@comcast.com> wrote: > +1 TSFatal() > > On 2/16/18, 2:25 PM, "Bryan Call" <bc...@apache.org> wrote: > > +1 on TSFatal(). We already have Fatal() in the code. > > -Bryan > > > > On Feb 16, 2018, at 7:33 AM, Alan Carroll > <solidwallofc...@oath.com.INVALID> > wrote: > > > > PR 3129 was a fix to the ICC builds that disables a "expression is > > constant" error which resulted from code that looks like > > > > TSReleaseAssert(!"error message") > > > > The goal of this code is to (1) force a process shut down and (2) > with a > > specific error message. We discussed some alternatives and the one I > liked > > best is to add a TS API function something like "TSPanic" or > "TSFatal" > > which works like the assert but always aborts. This might make the > code / > > intention a bit clearer as not everyone is familiar with this > particular > > paradigm. Any suggestions or comments on the naming of this > function? I > > prefer "TSFatal" myself. > > > >
