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

Re: New filesystem standard - do we want it ?



Lalo Martins wrote:
> 
> Maybe we could come up with a "transition path" - like moving
> stuff like /usr/doc (less prone to make the system break) and
> then symlinking "ln -s /usr/share/doc /usr/doc"
> 

Yes, make this symlink (ln -s ../share/doc /usr/doc) on freshly
installed systems (where /usr/share/docs will really exist), while 
make the opposite (ln -s ../../doc /usr/share/doc) on existing systems 
(where /usr/doc is still there).

We should create a compatibility package that installs those symlinks
(the idea can be used also for other paths) and put it on hamm. Then we
should upgrade the version-package (say to 2.0) and have dpkg
automatically add a dependency on this compatibility package if the
version is lower (what do you think, could this be done, Ian?).

This way we can have a long transition time where packages that install
in the old path and packages that install in the new path can cohexist
without none noticing the difference.
When all packages have been changed, then we can remove the symlink and
(in the latter case) rename the dir.

FYI, this idea isn't mine, since I've seen it used by IBM to handle the
transition between the (broken) file system structure used in AIXv2 to
the current in AIXv4: during 5 years (all AIXv3 lifetime) the symlinks
were part of the core system, while in AIXv4 they are in a
compatibility package that the user must decide to install if needed.
People didn't noticed this untill the symlinks disappeared (and then
complained :-) and this means to me that the system works.
(AIX doesn't have a clever dependency system; we have, so we can do
better :-).


Cheers,
Fabrizio
-- 
| fpolacco@icenet.fi    fpolacco@debian.org    fpolacco@pluto.linux.it
| Pluto Leader - Debian Developer & Happy Debian 1.3.1 User - vi-holic
| 6F7267F5 fingerprint 57 16 C4 ED C9 86 40 7B 1A 69 A1 66 EC FB D2 5E
> Just because Red Hat do it doesn't mean it's a good idea. [Ian J.]



Reply to: