summaryrefslogtreecommitdiff
path: root/NOTES
blob: 73262c0ea2f3f31c3c36cd5412c15c773111826c (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
This is the Guile developer notice-board.  -*- change-log -*-
----------------------------------------------------------------------
CONTENTS: Notes of various kinds which the Guile developers want to
share among eachother, e. g., a memo about something which needs
fixing.  Describing a certain problem here means that everyone is free
to fix it when and as he wishes.  (Of course it can sometimes be suitable
to discuss it first.)

RULES: Entries should have a date and the name of the author.  Entries
should be sufficiently detailed to enable other members of the team to
understand them, but they need not be comprehensible to people who
don't spend much time on Guile.  This file should not go into
snapshots or distributions, but is a strictly internal document.

CONVENTIONS: ChangeLog format. (Reverse chronological order.)
----------------------------------------------------------------------

Sun Mar  9 15:45:14 1997  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* We need to discuss the long-term solution to Guile configuration
	and customization.

	* Some day when more important stuff (module system, threads, Tk
	interface etc) has been completed, we should split up SCM's
	ChangeLog among us, and go through all changes since the birth of
	Guile.  We might find many nice bug fixes and improvements...

	* libguile/gc.c (scm_gc_sweep): The probability of collecting a
	free cell should be very low (only occurring when a signal has
	interrupted allocation).  Nevertheless, the GC immediately exits
	when the free cell test is enabled in the sweeper.  Where does
	this free cell come from?

	* libguile/arbiters.c: Remove this file?  When we've implemented
	POSIX thread support, "arbiters" will be superfluous.

Thu Mar  6 00:54:59 1997  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* libguile/unif.c: Introduce "fancy printing" for vectors.  "Fancy
 	printing" is performed by the printer when the `fancyp' flag is
 	set in the print state (see print.h).  One of its effects is that
 	sequences should be printed with maximum `length' objects.  (This
 	is currently only used in backtraces and error messages).
	This works for lists, but not yet for vectors and arrays.

Wed Mar  5 22:56:19 1997  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* libguile/struct.c, libguile/struct.h, libguile/gc.c: Structs
	need reimplementation.  Both user interface and representation
	need to be improved.  E. g., the struct part of scm_gc_mark need
	to be a lot more efficient.  We should probably do this redesign
	when designing(/porting) the object system.

	* libguile/symbols.c (SCM_SYMBOL_HASH): This is slot is currently
 	not used for anything (except that it is used by symbol-hash).  I
 	suppose it was originally intended to avoid doing multiple hashing
 	steps when handling symbols.

Mon Mar  3 21:36:58 1997  Mikael Djurfeldt  <mdj@mdj.nada.kth.se>

	* libguile/procprop.c: The current implementation of procedure
 	properties of closures uses a special property slot
	(SCM_PROCPROPS (closure)), but primitive procedures don't have
	such a slot.  Setting or getting source properties for primitive
	procedures currently involves making fake closures.  These are
	stored in an alist => finding the properties of a primitive
	procedure is O (n primitive procedures with properties) which is
	unacceptable.