"Juergen A. Erhard" <[EMAIL PROTECTED]> writes:
> Package: task-python-dev
> Version: N/A;
> Severity: important
^^^^^^^^^
Really? Is this really release-critical? Personally, I think
_closed_ would be a more appropriate state for this non-bug.
> There are too many Depends... most of those should be Suggests:,
> and some of them should be dropped completely.
>
> Stuff like python-gtk is okay, I think.
Three GUI libraries are fine, but two SQL interfaces are too many?
> Stuff I would move to suggestions:
>
> gadfly: if I use PostgreSQL, I have no use for this...
task-python-dev is supposed to be a catch-all package, a way of saying
"please give me everything a Python programmer might use, I'll sort
through it later". It's not about what you have a use for, or what I
have a use for, or what the maintainer has a use for, it's about what
constitutes "everything Python".
Task packages will never satisfy any particular individual completely;
they are a way of rapidly getting a lot of useful stuff, with the
tradeoff that you get some useless stuff as well. We can't tell in
advance what is "useful" and "useless" to any particular person in
advance. For example, most Python developers probably won't use three
different GUI libraries, yet there they are in task-python-dev...
because we don't know which one users want. If you know what specific
packages you need, task-python-dev is not for you.
In the specific case of gadfly, the existence of python-pygresql in
the task dependency list does not mean that there is actually a
PostgreSQL server there for it to connect to. (The dependencies of
python-pygresql will not cause the rest of PostgreSQL to be
installed.) Gadfly provides a lightweight SQL system that can be used
immediately by everyone, without installing additional software.
In fact, Gadfly can be very useful even if you DO have a full
PostgreSQL installation... personally, I have both PostgreSQL and
MySQL running here (my job involves Python/DBMS connectivity) and I
still use Gadfly, just because it's fast and simple for small things.
There are probably many things which can benefit from being able to
treat their data like an SQL database, without actually using one...
Another important consideration is that Gadfly follows the Python
DB-API, and last time I checked PyGreSQL still does not. If we remove
Gadfly, the only SQL interface provided by task-python-dev will be a
non-standard one. If you are developing database code in Python and
you want to have any hope at all of easily supporting more than one
SQL server, targeting the DB-API is a good idea... but if we remove
gadfly, the only Python/SQL interface provided in task-python-dev will
be a non-standard one.
I won't comment on the other specific packages you list, because the
argument is the same for all of them: you may not personally have a
use for them, but I could say the same about some stuff that you
probably DO use. If we are going to start arguing about what is
useful or useless, we might as well drop task packages altogether and
go back to making everyone scroll through the list of stuff in
dselect...
> I think all the task-python* packages lack focus...
They do? task-python and task-python-bundle are just small and full
versions of the basic Python install; if you think they lack focus,
ask Guido to stop putting so much stuff into the Python library.
task-python-web seems like a rather obvious one, and it seems to be
right on target. So the only one at issue is task-python-dev, which
is explicitly a catch-all.
Maybe in woody we can have some more task-python-* packages for more
kinds of Python programming... task-python-scientific and
task-python-sql might be good ones, for example. But for now, the
current ones are fine.
--Rob
--
Rob Tillotson N9MTB <[EMAIL PROTECTED]>
