Re: Packaging ACT-R -> Answers from upstream
On Thursday 19 May 2005 04:47, Ben Finney wrote:
| On 18-May-2005, François-Denis Gonthier wrote:
| > I would like to know what I should do about this email I received
| > from the maintainers of a upstream package I would like to package.
|
| <snipped>
| On the other hand, they do seem to be directly addressing the question
| of support for re-packaging (with a "no sir, we don't like it").
I think you are misreading Dan Bothell's reply here.
He wrote:
% However, because we often update the system,
% to prevent any issues with version differences
% we do not support any distributions other than
% the ones obtained from our web site
% and prefer that people only acquire the software directly from us.
To me this sounds like: the reason he doesn't like indirect distirbution
is that it adds a delay.
This seems a valid concern, even if Debian would release every year.
So either Debian ships obsolete packages in stable,
or the packages only exist in unstable,
or they don't get into Debian at all.
| > I just think users would benefit from having this package integrated
| > with their favorite common lisp environment. Getting that software
| > to work has been quite painful to me without a Debian package.
If that is what you are after, then you have a much better alternative.
Package as usual, put the debs on the web,
tell other users to put that in their sources.list .
This also gives you reliable info on how many users use it,
and thus how desirable it would be to distribute it through Debian.
| > I also suggested him that it was doable to build a package that
| > builds a package using freshly downloaded sources.
| Do-able, but hardly desirable.
Also, building is not the same as distributing.
Unless you distribute it from your own website.
Or put a package into Debian that is never meant to go into stable.
| Much better to help both the end-users *and* the upstream authors, by
| managing bug reports and feature requests, fixing them for your users
| if possible, and establishing an ongoing relationship with the
| upstream authors to get those changes integrated.
Yes.
This would make life easier for other Debian users that use this software
(if any),
plus that it would give ACT-R users more time to work on ACT-R,
instead of needing to spend it on getting the software to work.
After all, if you, a debian maintainer, found it difficult,
then for the average research psychologist it may take months.
It may even make ACT-R available to a wider audience.
This does not mean that Dan Bothell is not a capable programmer ;
distributing is just a specialized field.
You could add a lot of value to ACT-R,
for example automatically providing users with updates to the documentation
(using a package that downloads the docs,
and increasing the version of that package when the docs change)
Currently ACT-R seems to be a fairly young project,
thus the changes to it can impact it so severely that
it is not desirable for anyone to use anything but the latest version.
In time this may change.
Then the advantages of getting it into Debian might become available to it :
- automatically available on 13 different architectures,
including supercomputers,
- whole infrastructure for translations in place,
so less chance of translated versions starting to lead their own life
- vetted by the security team to prevent it being the cause of damage to
computers it is installed on (such things happen!),
- maybe even a game with which users can compare their own intelligence
with the intelligence of an ACT-R model
(should let the user win most of the time ofcourse,
so they would like to enable automatic sending of results to upstream)
such a game is much easier to create on Debian than on
a proprietary system, as Debian comes complete with everything.
please feel free to quote this message.
i read -mentors, so no need to CC me.
have fun !
Siward
(home.wanadoo.nl/siward)
Reply to: