bug-gawk
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Avoid gawkextlib as a separate shared library


From: Manuel Collado
Subject: Re: Avoid gawkextlib as a separate shared library
Date: Wed, 18 May 2022 11:11:13 +0200
User-agent: Mozilla/5.0 (Windows NT 10.0; WOW64; rv:38.0) Gecko/20100101 Thunderbird/38.5.0

El 17/05/2022 a las 18:43, Andrew J. Schorr escribió:
On Tue, May 17, 2022 at 08:54:01AM -0600, arnold@skeeve.com wrote:
Manuel Collado <mcollado2011@gmail.com> wrote:

IIRC, there have been other people confused about the role of gawkexlib
as a required DLL. ....
[...]
Is it ok to discuss the issue again?

I don't object. I think anything that simplifies gawkextlib installation
is good.

I don't mind discussing it, but I do mind implementing it. We have a working
solution that even MinGW libtool handles properly. With what few spare cycles
I have, I'd rather work on building a front-end installation script than
reengineering the existing infrastructure.

Ok. Please let's do a bit of discussion. Just to clarify ideas.

My point is to simplify building gawkextlib extensions from source distribution tarballs. If they need libgawkextlib-0.dll. Possible alternatives:

(1) An installer script or package manager/builder that takes care of dependencies. Please consider that a MinGW version is also necessary.

(2) Include /lib in the tarball of every extension that depends on it. Let 'make' build both the extensions and the gawkextlib base library. Let 'make install' copy gawkextlib files only if not already installed. Or better, check the version of the already installed ones and overwrite them if older.

(3) Consider varinit.c and the related macros in gawkextlib.h as an enhancement of the gawk extension API. Move those pieces to the gawk core. Handle strhash.c as any other source file required by a given extension.

(4) Handle the /lib stuff as regular source files required by specific extensions, and link the objects statically. My original suggestion.

Does any of those alternatives make sense? How can they compare w.r.t. the effort required to implement?

Regards.

--
Manuel Collado - http://mcollado.z15.es



reply via email to

[Prev in Thread] Current Thread [Next in Thread]