There might be hidden GC problem in ext/pg.c about ruby-pg HOT 4 CLOSED

Original comment by Michael Granger (Bitbucket: ged, GitHub: ged).

Thanks for bringing this to my attention! I'll familiarize myself with the problem a bit more and then apply the fix.

from ruby-pg.

Original comment by Anonymous.

Perhaps rb_ary_new2() is calling malloc() for RARRAY(ary)->ptr.
malloc() does not bzero() the buffer on most platforms.
I'd say it would be a bug in MRI rb_ary_new2().
When you call Array.new(100) you get an Array with 100 nils.

from ruby-pg.

Original comment by Michael Granger (Bitbucket: ged, GitHub: ged).

I'll look at this for 0.11.0.

from ruby-pg.

Original comment by Michael Granger (Bitbucket: ged, GitHub: ged).

Avoid use of uninitialized Arrays (fixes #47).

Instead of declaring an Array at the beginning of fetching results and pushing values onto it to return at the end (which could result in a segfault if the GC runs during a NEWOBJ), collect an array of VALUEs on the stack and create the Array on return with rb_ary_new4().

from ruby-pg.