On Jul 25, 2007, at 4:25 PM, Ralf S. Engelschall wrote:
On Wed, Jul 25, 2007, Jeff Johnson wrote:
On Jul 25, 2007, at 3:59 PM, Ralf S. Engelschall wrote:
Ralf: We again are at cross purposes, sigh. You know where I
would do the
dirty ...
I'm just fine with such a, say tool/db_tool.c, which integrates
all the db_xxx.c into a single tool. Let me investigate, perhaps
I can hack together this for us...
Ok, I've quickly done a feasibility study. See the attached patch, it
integrates the tools. BUT, big problem: the db_printlog and db_svc
tools cannot be easily integrated as they depend on a larger set of
additional DB object files and if we want to merge those two tools
also
into db_tool one would have to do a nasty replication of information
from db3/Makefile to tools/Makefile or we have to do the old nasty
"make
listobjs" stuff. Hmmm... I'm still not sure whether all this hacking
is really worth the efforts. Especially, as at the end the hacking is
larger than the benefit IMHO...
We needed the answer for Those Who Cry Bloat!. We seem to have
such an answer ;-)
If we can also find an answer for -Wl-as-needed users, then we can
proceed to the real issue at hand, partial static linking with -lrpm
"stuff" in order to start uncoupling API from ABI issues for rpm-python
bindings and more, thereby increasing reliability robustly.
Make sense?
73 de Jeff
______________________________________________________________________
RPM Package Manager http://rpm5.org
Developer Communication List rpm-devel@rpm5.org
