GMP and assert()
GMP and assert()
Posted Feb 28, 2019 15:50 UTC (Thu) by intgr (subscriber, #39733)In reply to: GMP and assert() by NAR
Parent article: GMP and assert()
There are some frameworks that handle assertion failures by logging and returning: GNOME things like glib, GTK, etc.
Have you ever tried running a GNOME application in a terminal? The majority of them spew errors and warnings left and right; developers seem to just ignore the errors because they are so easy to ignore. No-one bothers reporting them as bugs, although many of them probably are.
As far as I can tell, the whole industry has been slowly moving towards a more "fail fast, fail loud" approach. I'm surprised that it has suddenly became controversial.
Posted Feb 28, 2019 16:41 UTC (Thu)
by NAR (subscriber, #1313)
[Link] (3 responses)
Posted Feb 28, 2019 20:15 UTC (Thu)
by mageta (subscriber, #89696)
[Link] (1 responses)
Posted Mar 3, 2019 13:41 UTC (Sun)
by hadess (subscriber, #24252)
[Link]
There are 2 type of "errors" that crash the entire process in a GLib program: a call to g_error(), and memory allocation failure. Both of those are documented.
Posted Mar 1, 2019 18:31 UTC (Fri)
by notriddle (subscriber, #130608)
[Link]
Posted Mar 3, 2019 13:45 UTC (Sun)
by hadess (subscriber, #24252)
[Link]
They're easy to ignore because most "GNOME applications" aren't launched from a terminal and people rarely read all the errors in the journal when apps otherwise function.
And application developers are thankful for the bug reports, when they happen.
Note that if you feel courageous, you could set "G_DEBUG=fatal_warnings" and all the warnings would make the app crash. Send backtraces!
GMP and assert()
GMP and assert()
GMP and assert()
GMP and assert()
GMP and assert()
