In message <[EMAIL PROTECTED]>, Justin Erenkrantz writes:
>yet.  But, it does mean that CVS has a very short life expectancy within the 
>ASF.  So, expect one day that infrastructure@ will announce, "No more CVS 
>after X/YY/ZZZZ.  All ASF projects will be migrating to Subversion in X 

Excellent.  I never understand when programmers, who do something as
complicated (to most people) as programming, can't pick up or resist
picking up another tool.  We learn to program in languages we've never
programmed in before in under a day, yet it's hard for us to make the
switch to Subversionfrom CVS?  It doesn't make sense to me.  Subversion
has a lot of documentation and tries to stick to CVS command names and
flags as much as possible.  And I say that as someone who doesn't use
svn day in and day out.  Personally, I'm waiting for that 1.0 release
so I don't have to do dumps/loads every time there's a change in the
repository schema.

On the product group topic, I think jakarta-oro and jakarta-regexp
would benefit greatly from a move to commons.apache.org, but I've
not initiated that discussion because I haven't found the time to
follow through on the last topic I brought up concerning those
projects on the Jakarta PMC list.  As far as categorization goes,
since Commons houses reusable libraries and components, I don't
see any reason why components can't belong to multiple categories.
If each component has a set of properties, including name, language(s),
type(s) of functionality, binary size, etc., then we can autogenerate
a set of static Web pages from the component database (could be
a an XML file) that allow prospective users to find components based
on their requirements.  For example, someone working on a project in
C who needs an http client would start by choosing
Language.C->Category.Networking->SubCategory.HTTP to find Serf.
Someone else might choose
Category.Networking->SubCategory.HTTP->Language.Java to find HttpClient
Someone with less specific requirements might just look at then
entire component list sorted by name and browse the descriptions.  I
don't see a need for a single taxonomy.  That said, we still have
to identify the dimensions and categories to build the component
database.  I expect the idea is it will never get so large that we
would need to support dynamic queries because of ongoing subproject
promotion to top-level status.

daniel


Reply via email to