Mailing-List: contact zsh-workers-help@zsh.org; run by ezmlm
Precedence: bulk
From: Bart Schaefer <schaefer@brasslantern.com>
Message-Id: <160124102244.ZM11864@torch.brasslantern.com>
Date: Sun, 24 Jan 2016 10:22:44 -0800
In-Reply-To: <F46CF6DC-AC2E-4C64-B94E-E48B3FB68141@kba.biglobe.ne.jp>
Comments: In reply to "Jun T." <takimoto-j@kba.biglobe.ne.jp>
        "V01zmodload fails on Cygwin" (Jan 24,  8:10pm)
References: <F46CF6DC-AC2E-4C64-B94E-E48B3FB68141@kba.biglobe.ne.jp>
To: "zsh-workers@zsh.org" <zsh-workers@zsh.org>
Subject: Re: V01zmodload fails on Cygwin
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii

On Jan 24,  8:10pm, Jun T. wrote:
}
} V01zmodload.ztst fails on my Cygwin (on 64 bit Windows7):
} CYGWIN_NT-6.1 2.4.0(0.293/5/3) 2016-01-15 16:16 x86_64
} (probably the latest 64bit cygwin)
} 
} So it seems some symbols from libzsh-5.2-dev-1.dll also exists
} in param_private.dll, and gethashnode2 is one of those symbols:

Interesting.  gethashnode2 should be declared external, so it should
be resolved for param_private.dll at link time.  Seems to have to do
with whether the external function is *called* vs. merely referenced
by name?

} The following patch seems to work, but I hope someone who knows
} cygwin better to confirm/improve the patch.

It's too bad the node functions are direct members of the hashtable
struct rather than encapsulated the way the parameter GSUs are.  In
the absence of an entire data structure to save/restore, your patch
looks fine.  Should probably save/restore ->getnode as well, just
for completeness.

Do you want to commit or should I?