On 2018/09/13 12:21, Pierre-Emmanuel Andre wrote:
> On Thu, Sep 13, 2018 at 06:51:16AM +0200, Rafael Sadowski wrote:
> > On Mon Sep 10, 2018 at 03:10:09PM +0200, Pierre-Emmanuel Andre wrote:
> > > On Sat, Sep 08, 2018 at 02:20:49PM -0500, Matthew Martin wrote:
> > > > On Sat, Sep 08, 2018 at 08:36:29AM +0200, Rafael Sadowski wrote:
> > > > > Hi All.
> > > > >
> > > > > please find below a simple diff to update ZSH to the latest stable
> > > > > version. This release includes security fixes: CVE-2018-0502,
> > > > > CVE-2018-13259.
> > > >
> > > > zsh 5.6.1 will be out soon (it's tagged and built; just not announced).
> > > > It fixes the error which (I suspect) is why you switched to gmake.
> > > >
> > > > > Notable port changes:
> > > > >
> > > > > - Use configure elements instead of post-install tasks.
> > > > > -- Avoid version in path name. I don't see any reasons for that, I
> > > > > only
> > > > > provide one ZSH version.
> > > >
> > > > This is the upstream default, and I don't see a compelling reason to
> > > > deviate from it.
> > > >
> > > > Patch for 5.6.1 below. It was rolled a bit over an hour ago, so not well
> > > > tested yet. Obviously we should wait until it's officially announced.
> > > > There's one test error over what 5.5.1 had which I haven't had time to
> > > > look at yet.
> > > >
> > >
> > >
> > > I agree with Matthew. I prefer to wait for 5.6.1.
> > >
> >
> > I also agree with Matthew. Thank you for your feedback, Matthew.
> >
> > Matthew's and my diff below. Feedback? Ok?
> >
>
>
> Thanks. There is a regress failure:
>
> ./A05execution.ztst: starting.
> Pattern match failed, line 1:
> -<*tstcmd-interp-too-long: bad interpreter: x*xn: no such file or directory
> <status 127
> +>(eval):2:
> /usr/ports/pobj/zsh-5.6.1/zsh-5.6.1/Test/command.tmp//tstcmd-interp-too-long:
> bad interpreter:
> xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx$
> xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxn: exec format error
> >status 126
> Test ./A05execution.ztst failed: output differs from expected as shown
> above for:
> path=($shpath $echopath ${ZTST_testdir}/command.tmp/)
> tstcmd-interp-too-long 2>&1; echo "status $?"
> path=($storepath)
> Was testing: path (5)
> ./A05execution.ztst: test failed.
>
>
> This test success with the 5.5.1.
> I will look at it.
>
This test is looking for an OS dependent string (and the test changed
between 5.5.1 and 5.6, and again between 5.6 and 5.6.1). I think it's
just a slightly bogus test and can be ignored for ports.
