Re: [boost] Header dependency table

From: Beman Dawes (bdawes_at_[hidden])
Date: 2002-08-07 19:41:35

At 05:51 PM 8/7/2002, Douglas Gregor wrote:

 >I've managed to shrink the file size down to something more reasonable (< 

 >200KB). If you're still interested in checking it out, it's at the same
 >place:
 >  http://www.cs.rpi.edu/~gregod/hdr_depend/index.html

Definitely, 100%, cool!

But the effect it has is to make me even surer we are barking up the wrong 
tree when we concentrate on headers.

Step back a bit.  Look at the big picture.  What question are we trying to 
answer for users?  What is the use scenario?

Nico Josuttis posed the original challenge.  It went something like this:

"Suppose I'm considering using a Boost library. What other libraries will 
also have to be installed because the lib I'm interested in is dependent on 
them?

That's the question we need to be able to answer, and it seems library 
rather than header oriented.

Can you think of a use where knowledge at the header level is required 
rather than the library level?

 >> In other words, the granularity desired "depends on" column is really 
at
 >> the library than at the header level.
 >>
 >> Does that make sense?
 >
 >Yeah, but it's a maintenance nightmare. Then every header has to be
 >associated with a library somehow. We can use the directory naming
 >scheme to some extent, but I wonder how far that will go...

Look at the "Library" column in the new compiler status tables.  That is 
completely deduced by the program, including the link to the library's 
documentation.  It works pretty well except utility.

If we promoted the iterator stuff and maybe one or two other categories to 
be libraries in their own right, even utility wouldn't be so bad. And 
factoring out iterators has already been suggested for other reasons.

--Beman