Re: [PATCH] Changed assert to OSMO_ASSERT

On Sun, Mar 03, 2013 at 03:05:02PM +0000, Harald Welte wrote:

Hi,

Two more ideas:

• since it is an error condition, fprintf(stderr).
• it might make sense to also write it to the osmocom logging, using LOGL_FATAL / DLGLOBAL
• while at it, we might also print a backtrace by means of osmo_backtrace().

What do you think?

the current usage will be inside the regression/unit testing, e.g. not having a dependency on the logging framework is quite nice to have.

For osmo_backtrace and frpintf would be nice. I think the value of osmo_backtrace will be a bit limited as we will have no symbol name for the test function (but this could come from the fprintf).

cheers holger