Re: Bug#177285: ITP: oak -- Python DNS Server

[Lars Wirzenius <liw@iki.fi>]

su, 19-01-2003 kello 23:58, Sean 'Shaleh' Perry kirjoitti:
> On Sunday 19 January 2003 12:59, Stefano Zacchiroli wrote:
> >
> > IMHO the fact 'oak' is written in python needs not to be mentioned in
> > short description field, a user installas a name server for its external
> > functionalities (feel free to consider it a black-box approach), not for
> > the language it's written in.
> 
> actually a user is likely to care because they may not want the python depends 
> or runtime penalty just for a dns server. they may be shopping for a 
> replacement to bind and are looking at the available choices.

In that case, the user should look at the dependencies, yes? Just like
when they should look at the dependencies for, say, bind9, in order to
notice that bind9 depends on a bunch of library packages in addition to
libc6.

The runtime penalty would have to be measured. Just because something is
written in Python (or Perl, or Haskell, or whatever language) doesn't
mean it's slow, or even that it is slower than a similar program written
in C. Sometimes it is, of course, but often it isn't, and people
shouldn't be encouraged to make such decisions based on the
implementation language.

I don't wish to incite a language war (on this list). I don't wish to
encourage any language over others, or discourage use of any particular
language (again, not on this list). However, I oppose a requirement,
even an recommendation, to mention the implementation language in the
description, unless it has a real, major impact on the user, by way of
functionality, and that if the user is interested in the size a package
and its dependencies require, this is appropriately reported using other
fields.