Crash in duk_push_vsprintf (on x86_64) about duktape HOT 4 CLOSED

Thanks for pointing this out - I've missed this bug because it probably only manifests itself when the initial snprintf() attempt fails and we need to grow the buffer. This only happens if the string to be printed is long enough. The initial guess is (currently) min(256, strlen(format_string) + 16).

Not sure how to fix this yet so it would also be portable to C89 environments. It seems some projects are just using a plain assignment (copy = ap) in pre-C99 environments and it works in most (but not all) such environments. I'll probably wrap va_copy behind a macro so that porting hackery can be localized if it is necessary for some specific pre-C99 environment.

from duktape.

Master branch now contains a fix which seems to be verified by api-testcases/test-bug-push-sprintf-va-copy.c. There is a non-C99 replacement which uses a plain assignment which is far from ideal but should be enough for now.

Could you verify this fixes your specific issue too?

from duktape.

Will verify later today when i get home.

from duktape.

Works fine for all cases I had

from duktape.