On Fri, 2001-11-30 at 03:26, Greg Stein wrote:
> Holy crap, that is some major suckage. Allocate a heap buffer so that we can
> put a constant string into the output XML? Blearg.
> I guess the one saving grace is that it isn't within a loop, and fb->pool is
> about to be destroyed. But sheesh... what convolutions for an otherwise
> simple task :-(
This is going to sound rude and patronizing, especially since you
probably have at least as much experience as I do, but I don't have a
better way of saying it: I think you need to learn to care less about
extra operations like this. Often, uniformity of interface is more
important than making every function of a program use the few possible
number of operations. (Obviously, there are exceptions: when the extra
operations are causing a real, observed performance problem, or are
making parts of the code difficult to maintain, then something has to be
done.)
That said, given the conventions we've adopted, I strongly suspect the
function involved should be taking a const char * instead of an
svn_stringbuf_t *. (An XML attribute cannot contain 0 bytes, so
svn_string_t * doesn't buy us anything over const char *.) So my
disagreement is in your calling it "major suckage," not in the proper
course of action.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Oct 21 14:36:50 2006