Zsh Mailing List Archive
Messages sorted by: Reverse Date, Date, Thread, Author

Re: Cores almost on demand in patcompile()



On Thu, 20 Oct 2016 13:09:25 +0200
Sebastian Gniazdowski <sgniazdowski@xxxxxxxxx> wrote:
> Tested the change, no segfault, also when break is added to unmodified
> code. Also checked if the pattern still matches (made use of the
> prepared input1.txt, input2.txt) and it does, in both versions.

Thanks.

I think that *is* the correct fix --- we add PAT_PURES to the flag that
will later be used when we run the pattern, and in that case any of the
usual wordcode infrastructure is irrelevant because we do indeed treat
it just as a single (metafied :-() string.  So breaking out of the loop
immediately is the right thing to do.

pws



Messages sorted by: Reverse Date, Date, Thread, Author