Re: inconstancy with RS = "(\r?\n){2}"

[Alex fxmbsw7 Ratchev]

well its not 2 but in this no \r data case it was used to
if one would stack chars and on each new check for RS it would work,
.. so its some bug

On Sun, Jul 25, 2021 at 2:37 PM Alex fxmbsw7 Ratchev <fxmbsw7@gmail.com> wrote:
>
> i think the bug is in the pre math'ing of the length ( max ) to
> interpret or something
> which in this case should be 2, but as gawk reacts to 4 or 6 or something
>
> On Sun, Jul 25, 2021 at 2:32 PM Alex fxmbsw7 Ratchev <fxmbsw7@gmail.com> 
> wrote:
> >
> > hm where are the changes reflected
> > in the end i need it from a tcpserver spawn, no tty
> >
> > On Sun, Jul 25, 2021, 14:27 <arnold@skeeve.com> wrote:
> >>
> >> Alex fxmbsw7 Ratchev <fxmbsw7@gmail.com> wrote:
> >>
> >> > i noticed on the terminal variant it processes after couple of \ns more
> >> > but thats still not good enuff for tcp httpd connections
> >>
> >> I think this fixes the issue for terminals.
> >>
> >> Arnold
> >> ----------------------------------------
> >> diff --git a/io.c b/io.c
> >> index 91c94d9b..4b38f092 100644
> >> --- a/io.c
> >> +++ b/io.c
> >> @@ -3759,7 +3759,7 @@ again:
> >>           */
> >>
> >>         /* succession of tests is easier to trace in GDB. */
> >> -       if (RSre->maybe_long) {
> >> +       if (RSre->maybe_long && (iop->flag & IOP_IS_TTY) == 0) {
> >>                 char *matchend = iop->off + reend;
> >>
> >>                 if (iop->dataend - matchend < RS->stlen)