Re: [gentoo-dev] GLEP33: Eclass restructure

[Gustavo de Lama]

Please, excuse me!

This is my first mail to any Gentoo list and I should be introducing
myself as the perfect unknown I am but, instead of it, I feel as if,
with the simple mindedness of the just arrived, were about to say aloud
the emperor to be nude.

The matter is I cant but give my two cents and say: SLOT.

Again, I ask for your forgiveness in case this has already been
discussed here. I have tried, and not been able to find any trace of it.

To make my point, let me say all I know about software engineering:
There are small "maintenance" changes, mostly bug crunching and,
perhaps, security incidences, that get by with a make number (the one
after the second dot in the version code). Then, you can find
functionality or packaging changes that deserve a new release number
(the one to the right of the first dot), even when some companies prefer
to call them "service packages". And, when something needs real
revamping, then comes the version number, the one that comes before any
dot in the version code.

The first example to justify a version bump is breakage of backward
compatibility, so... IMHO, this restructure means a new version. And it
is for new versions that slots came to be. So that you can yet use and
maintain the old version, while the new one stabilizes and begins to
produce its rewards (It is not without perils that you reuse the
name-space with something that is truly a different application, even if
with a functionality akin the one before).

This way, you could perfectly use the old slot to unmerge what it
merged, and the very Portage should get rid of it when it were no more
necessary.

If, after a time, any package would depend on the new slot and were
incompatible with the old one, that should only mean that, anybody to
use it, had to renounce to the backward version at all. Then, you have
to take your chances: it exists the possibility of beginning a new
Portage tree free of packages that depend on the old version, just to be
sure that there is a clean installation and migration path... and I
think that only with the slot solution may you be sure of this without
the wood.

Well, to conclude, slots are a small headache and not even completely
well supported by Portage, but I think now is the moment to consider
using them and, if, because of it, you end with a better understanding
and support for them, so the better. Quality is hard to begin with but,
at last, it pays for itself.

Thank you for your indulgence.

Gustavo de Lama.

--
gentoo-dev@g.o mailing list