GooNews

Teasers mainly, but it gives a good hint that wheels are still turning. (Wheels within wheels as well!)

October 21, 2003

released 0_154 with C embedding facility.

February 04, 2003

installed chris doubles win32 vc port of 0_153.

January, 2003

release 0_153 available.

December 31, 2002

jrb's back on the case. working on a new release...

December 10, 2002

jrb's been working on end of semester deadlines and then is off on vacation until xmas. plan is to get new release out then with increased speed, rationalized design, bug fixes, tinycc backend, etc.

November 15, 2002

James got tinycc 0.9.13 to work with goo as a replacement dynamic compiler backend. it's much faster than gcc and dynamic loading. unfortunately, it's not as portable.

November 6, 2002

jrb's been working on sound wall installation. now back to goo.

October 9, 2002

rough version of "goo at eighteen months" is now available. comments appreciated.

October 7, 2002

jrb working on finishing integrating pretty printer back into sources and documenting. getting ready for another release.

September 30, 2002

vm now properly runs test suite.

September 27, 2002

jrb got property access optimization working.

September 26, 2002

jrb gets pretty printer working and is working on greater rationalization of object printing.

September 25, 2002

jrb flushed out dependency tracking design.

September 24, 2002

jrb is working on optimizing property access and inlining method calls. jrb is working on a dependency tracking system that will support the tracking of assumptions made while performing type-based optimizations.

September 18, 2002

james is working on quantities. jrb is working on tracking refs more carefully to slim down generated dynamic code.

September 17, 2002

jrb is working on simple function profiling. the hottest funs are in order: ==, +, <, not, ...

September 10, 2002

jrb is working on simple type-inference and optimizing dynamic compilation.

September 08, 2002

,vm-eval mode is working. pushing more and more code through the vm.

September 07, 2002

loop optimizations are in place successfully converting and inlining functions with nested and sequential loops. this not only greatly improves performance but also allows gprof to give more comprehensible information.

September 03, 2002:

v146 is released with source line info and most of visual c support on win32.

August 31, 2002:

jrb is working on merging back in chris double changes to get goo running with visual c on win32.

August 29, 2002:

checked into cvs is a version of goo whereby source lines are tracked through compilation, stored in functions, and displayed during backtraces. this will be made into a release shortly.

August 23, 2002:

v143 is released including a condition hierarchy, stack overflow checks, sub* and elt!, and the introduction of new experimental syntax for with [ ... ] for elt/sub/sub* and { ... } for fun. consult the v42 manual for more information.

August 22, 2002:

select() support has been added to Goo's net stuff. Miniature chat server in demos/chat.goo. (Andrew)

Jonathan has a first-pass at a VM up and running in Goo checked into CVS, or at least parts of it.

August 19, 2002:

Goo 0.142 is out as of sometime this past weekend. You would have known this a while ago if you were on the mailing list (or you already know this).

In other news, if you have any desired features (linguistically or library-wise), please throw them up on the WishList. Wishing doesn't make it so, but it lets us know what people are looking for.

August 11, 2002:

I done posted the following to the list:

Anonymous CVS (see info at Wiki) now has use/mangle support checked in. This is relevant for those of you who want to do things like:

(use/mangle tests/manglea a:)
(use/mangle tests/mangleb b:)
(a:func-a)
(a:func-b)
(b:func-a)
(b:func-b)

All it does is import everything exported by the module in question, but with the given prefix mangled on to the binding name.

For those who were desiring module:binding, this is equivalent to what you wanted, although it's usable for other things too. For example, you could also bind on "mail-" instead, or something like that.

This should help deal with the instinct of prefixing method names yourself, out of a fear of name collisions. (Eg, earlier I found myself prefixing "mail-" onto functions just because I was concerned about binding collisions with something more important.) Note that I am one of those people would ideally like a unified generic system without the constraints on the number of parameters, but this is a useful thing regardless, and very helpful in the short term.

IMPORTANT!

I've checked in new .c files and a new binary as well because the boot-strapping issues require you to either just build your image from the C files/use the binary, OR you must go through the boot-strapping process yourself to get yourself a happy image. Specifically, the use/mangle export in src/goo/boot.goo requires that the running image must have the use/mangle export already loaded. So to build it yourself, you must remove the export, build an image, then build a new image using the new image, with the xport re-exported. (It's possible there's a way around this, but I'm not one of the people who knows how to do it, if so.)

Andrew

PS-- You'll need to (use goo/boot) to get at this.

August 10, 2002:

Er, so there's been anoncvs access for about a week now. C-client access to IMAP is in preliminary stages, and hopefully soon in a more advanced dancing state. Preparations are underway to use the GNU MP infinite-precision integer library; <int> has been made abstract, with <prim-int> being the actual tagged class. Lots of shoes in the stew, as it were.

July 30, 2002:

Got some sockets going on. (Andrew)

July 29, 2002:

Goo now has a primitive MIDI layer that outputs to the ALSA sequencer API. (Andrew)

July 20, 2002:

We have the GTK bindings (a la SWIG) now being statically built into a goo image, instead of requiring dynamic linking. The major benefit is that goo modules (like the UI Layer, SamurUI), can now be statically compiled as well, making for one major development speed-up.

June 30, 2002:

Threads a la Jonathan. Check out the prelim docs at http://www.ai.mit.edu/~jrb/goo/manual.40/goomanual_41.html.

June 14, 2002:

Bug tracking a la Mantis at http://djundjun.ai.mit.edu/mantis/

June 13, 2002:

JB has a first pass at threads working. Multiple threads doing stuff without dying horribly. Synchronization and comm are up next.