/[movitz]/public_html/ChangeLog
ViewVC logotype

Contents of /public_html/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1.13 - (show annotations)
Fri Mar 7 23:39:33 2008 UTC (6 years, 1 month ago) by ffjeld
Branch: MAIN
CVS Tags: HEAD
Changes since 1.12: +5 -0 lines
*** empty log message ***
1 2008-03-08 Frode V. Fjeld <frodef@cs.uit.no>
2
3 * movitz/losp/muerte/more-macros.lisp: Implemented macro
4 destructuring-bind.
5
6 2008-02-25 Frode V. Fjeld <frodef@cs.uit.no>
7
8 * movitz.asd: Created an ASDF system definition.
9
10 * movitz/asm.lisp, movitz/asm-x86.lisp: Created new assembler and
11 disassembler that's less overengineered (the design goals of
12 ia-x86 were not originally to serve as an assembler). This speeds
13 up compiles considerably; on the order of twice as fast, and
14 reduces the overall footprint of the movitz system too.
15
16 2008-01-13 Frode Vatvedt Fjeld <frodef@cs.uit.no>
17
18 * movitz/losp/muerte/memref.lisp: Fixed (setf memref-int :type
19 :unsigned-byte32), which didn't really work at all I think. It
20 would write to the wrong address, or go into unbounded recursion.
21
22 2007-03-14 Frode Vatvedt Fjeld <frodef@cs.uit.no>
23
24 * movitz/losp/x86-pc/keyboard.lisp: Patch from Shawn Betts
25 improves the keyboard driver and adds dvorak support.
26
27 * movitz/losp/x86-pc/vga.lisp: Patch from Martin Bealby adds some
28 simple VGA graphics capabilities, and a neat splash image.
29
30 * movitz/bootblock.lisp: Added detection of floppy geometry in the
31 bootloader. This was prompted by the fact that Qemu appears to do
32 some weird auto-detection of the floppy geometry based on the
33 floppy image's size. However, it should also mean that movitz can
34 now be booted from a "real" 5 1/4' floppy should you ever find
35 one, and more importantly, probably also 2.88-MB floppies as
36 supported by bochs and qemu, and I think also some actual hardware
37 drives.
38
39 * movitz/compiler.lisp: Much improved keyword argument parsing. It
40 is now faster, smaller in code-size, and more (fully?) ANSI
41 compliant.
42
43 * doc/ChangeLog: Things are not moving quite as slowly as might be
44 suggested by the lack of activity in this file. The last few
45 weeks, there has been quite a bit of activity. I'll try once again
46 to remember to update this ChangeLog.
47
48 2005-12-02 Frode Vatvedt Fjeld <frodef@cs.uit.no>
49
50 * Uploaded a new los0.img.
51
52 2005-11-02 Frode Vatvedt Fjeld <frodef@cs.uit.no>
53
54 * Fixed some aspects of the REPL code so that it will run in the
55 default bare-bones environment that a new thread gets. This makes
56 it more convenient to fiddle with threading.
57
58 2005-10-31 Frode Vatvedt Fjeld <frodef@cs.uit.no>
59
60 * Fixed threading to work, in the sense that the REPL code will
61 now work reasonably in the dynamic environment that a fresh thread
62 supplies.
63
64 2005-07-21 Frode Vatvedt Fjeld <frodef@cs.uit.no>
65
66 * Fixed the compiler to handle e.g.
67 (defun foo (x &optional y)
68 (lambda () y))
69 which before would cause the compiler to fail with a message about
70 not being able to lend optionals.
71
72 2005-06-14 Frode Vatvedt Fjeld <frodef@cs.uit.no>
73
74 * Added dynamic growing and rehashing of hash-tables.
75
76 2005-06-11 Frode Vatvedt Fjeld <frodef@cs.uit.no>
77
78 * Started to add support for adjustable and displaced vectors.
79
80 2005-04-24 Frode Vatvedt Fjeld <frodef@cs.uit.no>
81
82 * Fixed a bug in the compilation of (values ...) for more than two
83 values: the stack-pointer would not be reset properly, causing the
84 stack to grow very quickly in certain situations, such as when
85 such a values statement was executed inside a loop, which happens
86 e.g. in do-symbols & friends. This bug has gone undetected because
87 usually a stack-frame exit will occur after a values statement,
88 masking out any failures.
89
90 2005-04-18 Frode Vatvedt Fjeld <frodef@cs.uit.no>
91
92 * Added support in the compiler for non-dynamic-extent &rest
93 function variables, so you don't _have_ to have those (declare
94 (dynamic-extent rest)) in there. Efficiency-wise, I'd still
95 recommend including them, though, because otherwise copy-list is
96 called to produce a infinite-extent list.
97
98 2005-03-09 Frode Vatvedt Fjeld <frodef@cs.uit.no>
99
100 * Added/much improved support for GC-migration of
101 code-vectors. Should now work well, except for some extreme corner
102 cases such as the interrupt-trampoline code-vector.
103
104 2005-02-24 Frode Vatvedt Fjeld <frodef@cs.uit.no>
105
106 * Entering an integer at the REPL now invokes the corresponding
107 restart, if it exists.
108
109 2005-01-29 Frode Vatvedt Fjeld <frodef@cs.uit.no>
110
111 * There's been quite a bit of work on scavenge.lisp. The stack scavenging
112 has been re-designed completely, and should now not break as it
113 sometimes did before in certain interrupt
114 situations. Interestingly, the scavenging API essentially didn't
115 change at all. There are still some holes left empty when it comes
116 to migrating code-vectors, but this should be trivial to fix.
117
118 2005-01-04 Frode Vatvedt Fjeld <frodef@cs.uit.no>
119
120 * Fixed some support for stack-allocating funobjs and cons-cells
121 of dynamic extent.
122
123 * New years resolution: Update this ChangeLog more often.
124
125 2004-07-19 Frode Vatvedt Fjeld <frodef@cs.uit.no>
126
127 * Bignums are now working to the extent that one can basically
128 expect things to work, although there still are arithmetic operators
129 without bignum support.
130
131 2004-07-13 Frode Vatvedt Fjeld <frodef@cs.uit.no>
132
133 * Added the feature that after each GC cycle, the keyboard is
134 polled. If the escape key is entered, a break to the debugger is
135 performed. This way one can interrupt long-running executions, so
136 long as the GC is triggered once in a while.
137
138 2004-07-12 Frode Vatvedt Fjeld <frodef@cs.uit.no>
139
140 * Added top-level-command :more, that will evaluate a form while
141 paging the console output. That is, prompt for a key-press after
142 each screenful.
143
144 2004-07-08 Frode Vatvedt Fjeld <frodef@cs.uit.no>
145
146 * Migration to the new basic-vector data-structure is completed
147 (though not thoroughly debugged). The los0-gc.lisp example GC is
148 updated to reflect this, and now supports any newspace-size upto
149 2GB.
150
151 2004-04-18 Frode Vatvedt Fjeld <frodef@cs.uit.no>
152
153 * The bootloader no longer initializes the VGA subssystem, since
154 the textmode driver in los0 should be able to adapt to most
155 standard VGA textmodes now.
156
157 * Apropos should work again, I think it didn't before due to a
158 compiler bug.
159
160 * There is now a fairly complete GC in los0, meaning in principle
161 there us no reason it should not keep running forever. I've had an
162 uptime of about a week. The new-space buffer is limited to 256 KB,
163 however.
164
165 2004-03-29 Frode Vatvedt Fjeld <frodef@cs.uit.no>
166
167 * Added basic GC functionality. The Movitz platform is now I
168 believe viable for at least some real-world applications! Also,
169 there have been many bug-fixes that I've forgotten to record here.
170
171 2004-01-19 Frode Vatvedt Fjeld <frodef@cs.uit.no>
172
173 * Fixed the bootloader slightly so it should now work with VMWare.
174
175 2004-01-12 Frode Vatvedt Fjeld <frodef@cs.uit.no>
176
177 * Fixed the floppy bootloader so that it will work with BIOSes
178 whose int13 messes with the GS segment register, like apparently
179 the Compaq and some others do.
180
181 2004-01-09 Frode Vatvedt Fjeld <frodef@cs.uit.no>
182
183 * A disclaimer has been added to the file COPYING.
184
185 2004-01-06 Frode Vatvedt Fjeld <frodef@cs.uit.no>
186
187 * Rewrote the floppy bootloader so that it is now only limited by
188 the size of floppies rather than the previous about 600 KB
189 real-mode buffer limit. For some time I've been forced to remove
190 some features from los0.img in order to squeeze it into 600 KB,
191 but no more. Bloat is bliss. However, booting with GRUB is always
192 an option, and probably the best one in the long run, and
193 certainly for stable installations (should there ever be
194 such). It's just that the floppy loader is very convenient during
195 development. [BB-ID: 1680]
196
197 2003-12-05 Frode Vatvedt Fjeld <frodef@cs.uit.no>
198
199 * Changed the way method dispatches are cached. It used to be
200 cached with hash-tables, now it's done with alists. The memory
201 usage and lookup-time should be improved for the common case,
202 although later on---when we have really big generic-functions---we
203 might want to be able to revert to hash-tables. [BB-ID: 1328]
204
205 2003-12-03 Frode Vatvedt Fjeld <frodef@cs.uit.no>
206
207 * :trace and :untrace are now implemented somewhat differently. In
208 particular, :trace takes one function-name, and a keyword argument
209 :callers that is a list of function-names, one of which must be
210 active (on the current call-stack) for trace to occur.
211
212 * There was a bug in the reporting of how much memory was being
213 consumed. That is, room and time would report double the actual
214 amounts. So now (time (cons 1 2)) should report 8 bytes.
215
216 2003-12-02 Frode Vatvedt Fjeld <frodef@cs.uit.no>
217
218 * Added some sort of simple multi-line capability for the
219 repl. It's not true multi-line editing, but at least you're not
220 limited to one screen-length of input any more.
221
222 2003-11-18 Frode Vatvedt Fjeld <frodef@cs.uit.no>
223
224 * Improved the style of interfacing hardware registers in the ne2k
225 driver. So now the driver is quite a bit smaller, and is again
226 included in the los0 image.
227
228 2003-11-10 Frode Vatvedt Fjeld <frodef@cs.uit.no>
229
230 * The last feature I added to eval was lambda the macro and the
231 keyword for the function special operator. So now lambda works as
232 one expects at the REPL.
233
234 * Started this ChangeLog now, as I hope to distribute Movitz more
235 actively. Not that this changelog will mostly be concerned with
236 user-visible changes. You can /not/ infer much about the level of
237 development activity from looking at this.
238

  ViewVC Help
Powered by ViewVC 1.1.5