Re: heads up
"Alfred M. Szmidt" <ams@kemisten.nu> writes:
> It would be easy to change /usr; we would need to have shadow
> translators, make existing packages install (which is trivial with
> the /usr->/ symlink), and so forth.
>
> We don't have a /usr -> / symlink anymore. It is optional, and the
> default is to create a seperate /usr directory, and only time you can
> choose to use it is when you install a new crosshurd which isn't done
> by many users.
Please keep track of the conversation and take note of what I say
before you reject it. We were discussing not what we have now, but
what we could do in Debian. And as I said, we need shadow
translators. Once we have them, we could create /usr->/ with little
fuss.
> What kind of redesign of the .deb format would you like? (You do
> know that .deb is just ar, right?)
>
> .deb is not "just ar". Removing pre/post install/uninstall scripts
> would be a good start since you can't run them if you are a
> translator. Then removing dpkg could be a nice second.
Let's change this around. What problem are you trying to solve?
Packaging formats are more than lists of files; saying "remove
maintainer scripts" because you want a pure list of files gets you to
a pure list of files, but how does it solve the problems which
maintainer scripts are there to solve?
> What sort of translator do you imagine for /share/info?
>
> /share/info/dir being a translator that always provides a updated dir
> entry for info readers to use. Put a info page into /share/info and
> viola, a entry in /share/info/dir pops up.
This would be easy to arrange in a Hurd-specific installer, and Debian
would be happy to include one.
> I think these are easy. You just give the right patch, and say
> "this provides a very nice feature on GNU/Hurd" and you're done.
> We have not had much problem in this regard.
>
> Sure, and that is why we have a nice warm flame war about /hurd each
> year right (although, always from people who have no clue about the
> issues involved)?
We haven't had one for years, actually.
Reply to: