[Nut-upsdev] megatec and some changes to cmdvartab and new-names.txt

[Carlos Rodrigues]

On 12/4/06, Arjen de Korte <nut+devel at de-korte.org> wrote:
> Carlos Rodrigues wrote:
>
> > The problem with the symlink approach is mostly a matter of driver
> > options in "ups.conf". I'm not sure "megatec" is fully compatible with
> > the other drivers in this respect, I'll have to check later when I get
> > home.
>
> That was my main worry too.

A quick "grep addvar" on the obsolete drivers shows that this can't be
done without adding some cruft to the megatec driver in order to
provide backward compatibility. I don't think there is something
significant to be gained by doing that.

> The only way to 'force' people to abandon obsolete drivers, is to break
> existing installations and remove them from the tree. I'm not ready for
> that in the (upcoming) stable release, however for the trunk I think the
> time has come to actively discourage the use by doing just that. If we
> provide (as a service) symlinks is something we might think about.
> Personally I would like to get as much input as possible from 'the
> field' about the new megatec driver. Therefor, removing the old drivers
> from the compatibility list would be a good thing for both stable
> release and trunk as well. New installations should no longer use them,
> existing ones should be supported until at least the next major release.
[snip]
> Sure. But (like you already mentioned) we should also try to persuade
> them to switch to the megatec driver. We could gently point them to the
> new drivers (testing/stable) or more forcibly (trunk). Some of the
> drivers it is replacing have been slowly rotting in the tree for years
> and I think that (at least for the development version in the trunk)
> they could be replaced.
>
> > While both drivers remain in the tree, maybe the normal banner can be
> > accompanied by an "upslogx(LOG_WARNING..."  telling that they're
> > obsolete. I guess that won't be redirected to /dev/null, although
> > users may not read the logs frequently enough and miss it anyways.
>
> That might be an idea, but likewise I have little hope that this is
> going to be read. Prepending a few lines in their respective man pages
> might also be a nice idea for the obsolete drivers (that won't hurt
> anyone either).

Ok. So, I'll do this (adding an upslogx to the driver, modifying the
manpage, and change the compatibility list to remove the obsolete
driver instances) for the trunk soon, and for testing in a few days or
something.

-- 
Carlos Rodrigues