Re: What is a crash?
David Kastrup <[email protected]>
| Newsgroups | gmane.comp.gnu.lilypond.devel |
|---|---|
| Organization | Organization?!? |
| Message-ID | <[email protected]> |
Dan Eble <[email protected]> writes: > The meaning of our "Crash" label is broad, for example, > > - segmentation fault > - out of memory > - user tired of waiting > - Scheme error > > I propose narrowing the definition of "Crash" to abnormal termination > by the OS and relabeling issues of the latter two kinds as "Defect". > > Does anyone see anything wrong with that? For me "Crash" means an abnormal termination not reported or at least differentiated by the application's error reporting layer (an application may catch segmentation violations and create a cute popup for them, but that would still be a crash in my book). That would include Scheme errors not occuring in a user-controlled layer. Followup errors by, say, a user creating callbacks returning the wrong kind of value (say) do not feel to be actually within that same scope. Instead of haggling over the definition of "Crash", maybe we should create explicit categories for the kinds of fifth wall breakage we want to see treated separately? -- David Kastrup