[elephant-cvs] CVS update: elephant/TODO
blee at common-lisp.net
blee at common-lisp.net
Sun Sep 19 17:41:43 UTC 2004
Update of /project/elephant/cvsroot/elephant
In directory common-lisp.net:/tmp/cvs-serv26883
Modified Files:
TODO
Log Message:
updates
Date: Sun Sep 19 19:41:43 2004
Author: blee
Index: elephant/TODO
diff -u elephant/TODO:1.5 elephant/TODO:1.6
--- elephant/TODO:1.5 Thu Sep 16 06:11:21 2004
+++ elephant/TODO Sun Sep 19 19:41:43 2004
@@ -8,35 +8,23 @@
tweak performance of transactions! dynamic-extent in CMUCL
/ SBCL.
-more documentation: reference!
+more documentation: texinfo NOTES.
-secondary index generation, cursors:
-
-- secondary indices on the "lisp" side : minor / nil
-performance gains and DB handle badness for DB->associate
-
-- create 2 DBs per sorting function: primary and secondary,
-without and with duplicates.
-
-- in addition to the usual lexicographic sorter, create a
-"lisp" version -- sorts primitive types (numbers, strings,
-symbols.) use http://oss.software.ibm.com/icu/ for 16-bit
-unicode.
-
-- equality joins have to be done on the lisp side:
+equality joins have to be done on the lisp side:
end-of-table is not the same as end-of-btree.
GC (need cursors)
Lispworks stuff (fli:replace-foreign-array...)
-bignum fix: CMUCL / SBCL use %bignum-ref. OpenMCL: check
-that ldb is non-consing (i think it is), look at
-%ldb-fixnum-from-bignum)
+bignum fix: OpenMCL: check that ldb is non-consing (i think
+it is), look at %ldb-fixnum-from-bignum. profile
+%bignum-ref on CMUCL / SBCL.
serialize lambdas, closures, packages.....this is hard!
-lispy pointer arithmetic (profile sap-alien, etc)
+openmcl lispy pointer arithmetic (profile sap-alien, etc).
+profile CMUCL / SBCL sap arithmetic.
performance hacks: class / slot to ID
@@ -53,4 +41,6 @@
incorporate requirements from ICU license
-cursor-put : move the cursor after insert.
\ No newline at end of file
+cursor-put : move the cursor after insert.
+
+change :transient flag to an allocation type (fix CMUCL!)
\ No newline at end of file
More information about the Elephant-cvs
mailing list