[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Derek R. Price
Thu, 01 Mar 2001 08:27:18 -0500
Pavel Roskin wrote:
> Hello, Derek!
> > 1) Distribute a stable release
> > b) If installed version >= packaged version, do not install
> It makes sence. I didn't know about that scheme. One important point here
> is that you will be unable to implant hacks into the included library,
> since chances are that it will be installed.
True, but linking against a shared library contatins the same dangers. In a
world, stable release of packages like zlib & libdiff would have APIs and a
bugs such that they wouldn't require hacking. Hence my opinion that if we have
change for those sections of the CVS code we should actually be talking to the
diff-utils &/or zlib developers.
> I don't understand what you mean, but probably we are talking about
> different things. If you are linking against libfoo and it depends on
No, I don't think we were.talking about different things. I meant that when a
Libtool lib gets installed that it installs a metadata file called libx.la
is package-dependant), then shoves the "real" libx.a & libx.so files in a
directory where they, in theory, won't be noticed.
You mentioned that use of libtool could spread like a virus because if the *.la
was installed then other packages needed libtool to use it. I was pointing out
if the *.a & *.so libs were installed in parallel rather than in the .libs
subdirectory then anybody could use the installed libraries.
I was wondering if I had simply missed a good reason that this hadn't been fixed
Derek Price CVS Solutions Architect ( http://CVSHome.org )
mailto:address@hidden OpenAvenue ( http://OpenAvenue.com )
Pi R squared?!?! Nooo! Pie R round, cornbread R square!
- Re: diff/Makefile.am,
Derek R. Price <=