On 04/27/2010 05:15 AM, Norbert Nemec wrote:
I am less worried about the size than about the quality.
Replacing any part of the standard library is an act similar to changing
the language definition, so including new modules into the library
should be done with as much care as including new features into the
language.
Understood. I believe that these libraries need to be maintained in the
same similar fashion as Phobos.
Apart from that, however, I like the "batteries-included" approach of
e.g. Python. Crypto, math, web services and many other parts are of
sufficient general interest to be part of the standard library. However,
if the quality of any module is questionable, it should rather be left
out until it is up to shape.
Understood. It's just that I have seen several complaints about
languages being "too hard to understand" due to the amount of libraries
that the language comes with. This was just one possible solution to
mitigate that especially since people are interested in contributing
code to Phobos.
Personally, I'm all for having lots of tools at my disposal at any time,
however I can see where there can be too much stuff in a standard
library. I believe that the Perl maintainers only add a new library to
the core language if it's heavily used, though that is not always the
case as I'm sure the DBI (the standard Perl database interface) is not
part of the standard library.
A CPAN-like repository with easy-to-retrieve additional modules is
certainly interesting as well, but it does not replace an extensive
standard library that you can rely on to be there whenever you write D
code.
Agreed, however sometimes it's just nice to be able to quickly find
something you need without scouring the bowels of the internet. It's a
big reason why I love Perl.
Casey
