summaryrefslogtreecommitdiff
path: root/NOTES
diff options
context:
space:
mode:
Diffstat (limited to 'NOTES')
-rw-r--r--NOTES70
1 files changed, 70 insertions, 0 deletions
diff --git a/NOTES b/NOTES
new file mode 100644
index 000000000..73262c0ea
--- /dev/null
+++ b/NOTES
@@ -0,0 +1,70 @@
+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.
+
+ \ No newline at end of file