Lambda the Ultimate

Re: lexical scoping:

And then I found out that my closures didn't capture the lexical scope. That pretty much ruined it for me.

I take it you find both Forth and Prolog unworthy for the same reason? Have you looked at upvar or uplevel? The mere fact that Guido finds them appalling means that they can't be all that bad, can they? They work very well in the context for which Tcl was originally designed.

Which reminds me... Although John Ousterhout's seminal paper³ has been mentioned in passing^4-9, it was never discussed on LtU as far as I can tell. A damn shame.

Anyhow. If you have a couple of hours to spare, you should go read The Tcl War. If you have a couple of days, you can read the entire thread.¹

On second thought, given that the original post by RMS was cross-posted to gnu.announce, gnu.utils.bug, gnu.misc.discuss, comp.lang.tcl, comp.lang.scheme, comp.windows.x.apps, comp.unix.misc, you're unlikely to find all of the follow-ups in this highly entertaining debate. RMS did not participate in any of the ensuing carnage but tried to "make amends" two years later.²

Footnotes

1. Why you should not use Tcl by Richard M. Stallman, Sep 24, 1994, Message-Id: <9409232314.AA29957@mole.gnu.ai.mit.edu>.
2. "Why you should not use Tcl" by "Richard M. Stallman"¹⁰, Apr 24, 1996.
3. Scripting: Higher-Level Programming for the 21st Century by John K. Ousterhout, IEEE Computer, March 1998, pp. 23-30. (A must read.)
4. 13 most cited PL articles, Sep 2004.
5. Two Languages (high+low level) vs One Language, LtU, Oct 2004.
6. Have scripting languages peaked?, LtU, Aug 2005.
7. The A-Z of Programming Languages, LtU, Jun 2006.
8. Purpose-Built Languages, LtU, May 2009.
9. on multiple fronts in 1994, LtU, Jan 2010.
10. Edited in response to comment-58302.

RMS did not participate in any of the ensuing carnage but tried to "make amends" two years later.

Did he? Read the last few lines of that post, and the ensuing discussion. It's not so clear who tried to make amends, AFAICS. I think there's someone here on LtU who might be able to shed a little light, though... ;)

It came out rather long so I stuck it on my own website rather than post it here:

The Infamous Tcl War

which begins:

On the blog Lambda the Ultimate Matt Hellige prompted me to explain a certain mystery surrounding the infamous "Tcl War" between the GNU project and Sun Microsystems. He noticed that some of the key "netnews" posts from Richard Stallman in the "War" had his name in the "From:" line, but my name in the signature. Here, I'll oblige by telling a story. Well, part of the story. I'm going to withhold some details because I'm just that kind of guy. In this story some innocents may have been named in order to protect the changed. Or something like that.

There are some mysteries I think are likely to remain permanent about the so-called Tcl war. "From:" lines, and signatures on certain famous messages are not necessarily a reliable guide to who wrote or posted what. It's a bit hard to describe how exactly whose names wound up where on what and probably best to not try. I'll just say that there was some half-friendly / half-hostile pranking involved.

I can offer some history, though - at least as well as I recall it 16 years later. The story I tell here is a bit long winded but includes some history about Java, Tcl, Javascript, and Guile.

Finally got around to reading this. Thanks for posting it. It's always very interesting to me to hear the inside view on events that occurred at a rather formative time for me, even though I was (a) much younger, (b) two thousand miles away and (c) decidedly an outsider looking in (well, I'm still that — I'm still two thousand miles away, too, now that I think about it).

And then I found out that my closures didn't capture the lexical scope. That pretty much ruined it for me.

I take it you find both Forth and Prolog unworthy for the same reason? Have you looked at upvar or uplevel? The mere fact that Guido finds them appalling means that they can't be all that bad, can they? They work very well in the context for which Tcl was originally designed.

I haven't used Forth ever. I've only used Prolog a bit, but never used higher-order functions or closures.

That one time I did have to deal with Tcl, the issue of not capturing the lexical environment came up when I reorganized the code to try and remove duplication ("Hey, Tcl has closures! This code can be way shorter!"). I did try searching for a fix, but I guess I was less familiar with the terminology back then. I wish I had stumbled across "upvar" back then so I wouldn't have had to undo my improvements.

I don't mean to bash the creators of Tcl or disregard the historical context. I'm just saying that, in a language whose standard library seemed to promote higher-order programming quite a bit, I was surprised that such simple cases didn't work "properly". And that's what ruined it for me.