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

Re: [D-I] Etch Debian Installer beta 1 released: i18n facts

Quoting Erdal Ronahi (erdal.ronahi@gmail.com):
> Hi,
>     Definitely, but it has been added to CVS trunk and fedora-branch, but not
>     to the glibc 2.3 branch, so it is not part of latest official tarballs
>     and some distributions will miss it.
> What does that mean? What is CVS trunk?

I guess that Denis is talking about the development tree ot the glibc.

So, the Kurdish locale has been added to the "trunk" of this tree,
from which releases are made.

The "fedora-branch" is, I guess, the branch from which Fedora's glibc
is released, while the "glibc 2.3" branch is the one from which the
"genuine" glibc is released.

Debian's glibc, and thus Debian's locales package, comes from the
genuine glibc branch. As a consequence, there is no Kurdish locale yet
in Debian's glibc.

We're again reaching a well known problem here:

-Debian locales come from the upstream glibc release
-Debian's glibc does not change too often
-upstream glibc does not change too often
-upstream is very conservative with locales

As an additional parameter, Debian specific patches to glibc are
minimized (lack of manpower, no real intent to implement new features
before upstream, I don't know)

As a consequence, the locales packages is very slow at reflecting
changes and epecially new locales.

This is why I now definitely intent to propose that D-I uses the
alternative, glibc-independent, locales packages by Denis.

This is not a proposal to abandon the locales package. We need to
continue using it for sending new locales or locales patches to
upstream glibc. 

But we also need more reactivity with locales. Definitely. As it does
not seem possible with genuine locales (probably for good reasons by
the package maintainers), we can do this another way.

Reply to: