Author: Philip Hazel Date: To: Ralf Junker CC: pcre-dev@exim.org Subject: Re: [pcre-dev] Regression: Lookbehind assertions with fixed number
of repetitions
On Tue, 15 Nov 2011, Ralf Junker wrote:
> Thanks for testing! I investigated further and realized that the problem
> only manifests if the PCRE_CASELESS compile option is set.
Aha! Yes, now I can reproduce it.
> case OP_EXACT:
> case OP_EXACTI: // Added this line - was it missing?
> branchlength += GET2(cc,1);
> cc += 4;
>
> Is my assumption correct? Could someone else test as well?
Without checking too deeply, that looks as if it is the fix. I will fix
it in the main code. In other parts of the library I've changed to
listing all codes explicitly, instead of relying on "default", so that
forgetting to add a new opcode is picked up when testing. I should do
the same in that code.