- From: Sam Varshavchik <mrsam@courier-mta.com>
- Date: Tue, 13 Dec 2005 18:54:35 -0500
- Cc: www-lib@w3.org
- Message-ID: <cone.1134518075.119325.12491.500@commodore.email-scan.com>
Vic Bancroft writes: > > Arthur Smith wrote: > >> >> Fewer library files (or just one) would be a little nicer - I've >> certainly had occasions where I accidentally left off one of the >> libraries and then was wondering what went wrong. No strong feelings >> on this though. >> >> Sam Varshavchik wrote: >> >>> I'd like to get a feel for what others think about this idea. If >>> there's interest then I can submit the patch -- probably after the >>> currently-pending code is released, so that I can resync against the >>> released version. >> > Yea ! This sort of work can proceed in two directions. It sounds like > you have nailed down the first. > > The second would be providing some aid in using just a part of the > library function and only pulling in the required parts . . . Well, a shared library is a shared library. It's a single hairball, an "all or none" deal. If you link against it, you're going to pull in the entire library at runtime. What you want to do is link against the static library, so that you only get what you need, and the flip side is that it's going to be a part of your executable, and not shared with anyone else. In this situation a single static library (as opposed to a dozen+ static libraries you have now) is even more convenient. You do not need to figure out all their interdependencies, and which ones you need to link against. Just one static library, link against it and you're going to get only the code you need.
Received on Tuesday, 13 December 2005 23:54:43 UTC