Re: gnupg problem
>From Hubert Chan on Thursday, 21 June, 2001:
>-----BEGIN PGP SIGNED MESSAGE-----
>>>>>> "Joseph" == Joseph Pingenot <firstname.lastname@example.org> writes:
>Joseph> So why not create in the virtual package an actual wrapper
>(well, because then it wouldn't be a _virtual_ package, but ignoring
Heh. How about "translucent" package? ;)
>Joseph> script (e.g. dpgpw for 'Debian PGP Wrapper') that detects the
>Joseph> version of pgp or gpg or whatever is installed on the system (or
>Joseph> chooses which one to use if multiple pgp implementations are
>Joseph> installed), detects which version is installed, and presents to
>Joseph> the end user a unified interface to the implementations (and
>Joseph> maybe allows the user to, through clever use of options, use
>Joseph> something that, say, pgp has but gnupg doesn't).
>See the package pgpgpg (which, by the way provides "pgp"). But it
>doesn't work all the time, as some packages really do require pgp. For
>example x-pgp-sig-el, at least the last time I tried it, wants to look
>in my ~/.pgp directory. Adding a symlink from ~/.pgp to ~/.gnupg didn't
>seem to help.
Hrm. I'll look at pgpgpg when I can. Thanks for the info. :)
The basic idea would be that, if you just need a generic pgp implementation,
use dpgpw and its unified interface. If you're somehow linking to a
*specific* pgp implementation, then you'd list that implementation instead
>Changing all the packages to work properly wouldn't be a simple task.
>(Not saying that it's a bad idea, though.)
Aside from the issues of creating a unified command line interface, I don't
see how. It'd require only writing a wrapper script of the form
dpgpw-<implementation> (which dpgpw would use to find compliant pgps) which
translates the unified commandline into the specific commandline. If you
don't want to participate in the unified interface, you don't have to; just
don't put out a dpgpw-* wrapper.
Now if we wanted to be *really* creative (and work a lot), maybe we could
also create a unified library interface. But that might be a *lot* of work,
and it'd have to be updated often. As it stands above, if a program uses
a pgp implementation's libraries, it'd have to list the implementation as
required, instead of our "generic" pgp interface.
Besides, now we're deeply into territory that I'm not familiar with (using
pgp-ish shared/static libraries), so I'm not sure that this is often an
issue; I don't know what uses such libraries or even if such libraries
"IBM were providing source code in the 1960's under similar terms.
VMS source code was available under limited licenses to customers
from the beginning. Microsoft are catching up with 1960."
--Alan Cox, http://www2.usermagnet.com/cox/index.html