home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!stanford.edu!agate!agate!usenet
- From: blojo@xcf.berkeley.edu (Jon Blow)
- Newsgroups: alt.lucid-emacs.help
- Subject: Re: GC idea
- Date: 10 Nov 1992 00:36:22 GMT
- Organization: Experimental Computing Facility, Univ. of California, Berkeley
- Lines: 26
- Message-ID: <asdglha3298ghsdgh-1@xcf>
- References: <9211060256.AA08035@thymus.synaptics> <9211060517.AA05701@thalidomide.lucid> <GRUNWALD.92Nov8212721@foobar.cs.colorado.edu>
- NNTP-Posting-Host: graft.berkeley.edu
- In-Reply-To: <GRUNWALD.92Nov8212721@foobar.cs.colorado.edu>
-
- Dick Grunwald writes:
-
- > Have others made efforts on this front before? If so, how did they
- > fare? This (better GC) would improve the user interface considerably,
-
- Just recently, I finished a generational copying collector for a language
- I'm doing. It is essentially a control language for a network server; it
- is driven by events that occur on TCP connections.
-
- Anyway, I was thinking about how convenient it would be to use the time-gaps
- between user input (when the program would just be waiting in select()) to
- do partial garbage-collection. I got pretty darn optimistic about the idea
- (I'm currently working on a fuzzy controller to predict approximately how long
- one can time-out to garbage collect without bugging users); but I was thinking
- about the technique in general and I realized that there was only one other
- program that I could think of that this was applicable to: emacs.
-
- If you garbage-collect partially, you can take small timeouts (100ms) to
- do significant portions of gc work and not upset things very much.
-
- Anyway, I will look at the internals of elisp to see how easy it would be
- to install a new gc. Heck, I might even try it.
-
- Has anyone else out there done this sort of thing?
-
- -Jon
-