[Top][All Lists]

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

Re: Automatically remove .libs directory

From: Ossama Othman
Subject: Re: Automatically remove .libs directory
Date: Fri, 9 Mar 2001 09:46:03 -0800
User-agent: Mutt/1.2.5i


On Fri, Mar 09, 2001 at 06:23:13PM +0100, Peter Eisentraut wrote:
> Robert Boehne writes:
> > I don't think that removing files during "make clean" is part of
> > libtool's job, that is what Makefiles are for.
> If it's libtool's job to create them, and libtool wants to be a
> transparent solution, then it is also libtools job to remove them.  Surely
> you can do it in the makefile, but in reality you can do everything that
> libtool does in a makefile, but it's nice that libtool encapsulates the
> details so I don't have to worry about them.

I agree with Peter, especially since I was the one that suggested
adding a "clean" option to Libtool. :-)

> > I realize you wouldn't necessarily have the name of this directory
> > as libtool might generate different names on different platforms,
> > but perhaps libtool could be patched to return this info instead?

The way I see it, Libtool should be able to "undo" what it has done.
This implies being able to erase files/directories that it has
created.  Suppose you're using libtool to generate a C++ shared
library (assume ML branch for now) with template instantiations in it,
then it is likely that the C++ compiler will generate a template
repository.  However, template repository names are
compiler-specific.  Some use files, while others use a directory with
files inside that directory.  How is automake to know which template
repository to erase?  Libtool already knows which compiler generated
the template repository so it can easily determine how to erase it.
You could certainly return this information from libtool, but then why
isn't the same thing done when building a shared library?  I guess it
all depends on how tighly coupled automake and libtool should be.

BTW, the template repository cleaning feature currently doesn't
exist.  It's something that I've been thinking about implementing. :-)

Ossama Othman <address@hidden>
Distributed Object Computing Laboratory, Univ. of California at Irvine
1024D/F7A394A8 - 84ED AA0B 1203 99E4 1068  70E6 5EB7 5E71 F7A3 94A8

reply via email to

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