Re: [Gnu-arch-users] Archives use cases

[Clark McGrew]

On Mon, 2003-10-06 at 23:22, Robert Anderson wrote:
> On Mon, 2003-10-06 at 10:32, Clark McGrew wrote:
> > Both of the cases are questions about how to manage the flat arch name
> > space when there are a large number of categories.
> 
> In an abstract sense, the flat name space is completely irrelevant to
> how you might handle a "large number of categories."

I'm reminded of the saying, "In theory, theory and practice are the
same.  In practice, they are different".  The practical question I'm
trying to grok is how to manage an archive with a large number of
categories without having brain lock.  Consider doing a "tla categories"
on a large archive.

> Beyond that, I'm not sure what the problem is.  Here's a fictional
> example of how I might organize things:
> 
> {TalksPapers} archive
> chicago03--talk--0--base-0
> chicago03--talk--0--patch-1
> ..
> chicago03--talk--0--patch-27 
> 
> Ok, I gave the talk, now I need to submit the paper.  I'll start with
> the talk, to cull some text and figures:
> 
> chicago03--talk--0--patch-27 => chicago03--paper--0-base-0
> 
> work, work.
> 
> Ok, next up: San Diego...
> 
> chicago03--talk--0--patch-27 => sandiego04--talk--0--base-0
> 
> Lather, rinse, and repeat until career over.

I see we're settling on similar patterns.  I've been very interested to
find out how others are solving the "publication problem"...  I checked
the wiki and FAQ looking for best practice recommendations, but didn't
find any advise.  Hopefully, this sort of knowledge can filter into the
wiki.  

> > How should I handle several hundred categories in a single archive.

> Can you please state what the problem is in specific terms?  You haven't
> identified any actual problem other than "there are lots of
> categories."  Why is that a problem?

For a practical example of the problems created by a large flat
namespace, on most unix systems you can check a directory listing of
/usr/lib.  On my system, there are about 1000 names.  Now, summarize the
libraries by topic and package.  

In essence, the problem is "lots of categories".  The solution might be
as simple as "don't do that..." and break the archive up by topic

For instance

address@hidden      for the low level support packages
address@hidden  for a tracking detector
address@hidden    for a different tracking detector
address@hidden  for muon range detector packages

I get the suspicion that you are approaching this from the "Can TLA
handle a large number of categories" point of view and I'm sure that is
only limited by the file system.  

I'm approaching the problem from an organizational point of view.  How
should we use TLA so that accidental naming collisions don't happen? 
How does a user browse all of the categories to find an interesting
one?  etc.

Here again, I hope the wiki will eventually contain recommendations on
how a large project might be structured.  But the first step is to
identify the "Right Way (tm)" to solve the problem.

-Cheers,
Clark

-- 
Clark McGrew                    Univ. at Stony Brook, Physics and Astronomy
<address@hidden>        631-632-8299