Re: pre-announce exmh-2.7.2
Brent Welch <[email protected]>
| Newsgroups | gmane.mail.exmh.devel |
|---|---|
| Message-ID | <[email protected]> |
Tcl's variable tracing facility is powerful, but it can also be heavy handed. In this case the flist array is being traced, so every update to any element fires (er, used to fire) a procedure call that was really only interested in a slight subset of the array. I was doing all this at the same time as the "array unset" optimizations that were also cleaning up code that looked through a huge array just to operate on a small part of it. So, different cases, but similar issues of looking through a large data structure when only a small fraction of it is interesting. >>>Chris Garrigues said: > > From: [email protected] > > Date: Thu, 06 Jan 2005 10:41:00 -0500 > > > > Yeah, trace can be really handy for that. The problem is that it ends > up being > > an awful lot like Intercal's 'come from' statement (the only other > actual > > implementation I've seen of it was the old IBM Fortran G compiler's > DEBUG > > stuff, which allowed stuff like 'AT 10' - though of course all the > Lisp-derived > > langages the whole *point* of the language is tacking that stuff onto > expre > > ssions > > with (cons (foo) (bar)) ;) > > Oh, oh, you intuited that I spent years as a Lisp hacker. I've been > exposed. > > Lisp doesn't actually have too much of this kind of thing, but if you've > ever > had the "pleasure" of looking at Prolog, it's nothing but this stuff. > > Chris > > -- > Chris Garrigues Trinsic Solutions > President 1611-B West 6th Street > Austin, TX 78703-5074 > > 512-322-0180 http://www.trinsics.com > > Would you rather proactively pay for > uptime or reactively pay for downtime? > > Trinsic Solutions > Your Proactive IT Management Partner > -- Brent Welch Software Architect, Panasas Inc Delivering the premier storage system for scalable Linux clusters www.panasas.com [email protected]