id	summary	reporter	owner	description	type	status	priority	milestone	component	version	resolution	keywords	cc	difficulty
818	string-hash error	Jim Ursetto	sjamaan	"This error has been causing failure in sql-de-lite, fps, and hashes tests on salmonella since 2012/01/06.  It succeeded on salmonella 2012/01/04.  Both runs are reported to be in Chicken 4.7.4 series but it is not known yet exactly which sha1 hash.

{{{
Error: (string-parse-start+end) Illegal substring START spec
#<procedure (string-hash s1126 . maybe-bound+start+end1127)>
#f
}}}

IRC transcript:

{{{
17:16:36 < zbigniew> http://tests.call-cc.org/master/linux/x86/2012/04/14/salmonella-report/test/sql-de-lite.html
17:18:21 < sjamaan> What exactly is it doing?
17:22:10 < zbigniew> hash-table-ref of a string=? hash table, probably via lru-cache
17:22:17 < zbigniew> failure beings here: http://tests.call-cc.org/master/linux/x86/2012/01/06/salmonella-report/tests/sql-de-lite.html
17:22:28 < zbigniew> begins
17:22:33 < zbigniew> ok here: http://tests.call-cc.org/master/linux/x86/2012/01/04/salmonella-report/tests/sql-de-lite.html
17:23:49 < sjamaan> Can't reproduce with a simple hash-table test
17:27:24 < zbigniew> other eggs are apparently having the same issue
17:27:25 < zbigniew> http://tests.call-cc.org/master/linux/x86/2012/04/14/salmonella-report/test/fps.html
17:28:03 < zbigniew> http://tests.call-cc.org/master/linux/x86/2012/04/14/salmonella-report/test/hashes.html (at end of tests)
17:28:28 < sjamaan> So this has been broken for months. Why didn't anyone report this?
17:29:07 < zbigniew> probably no one affected is using master
17:29:18 < zbigniew> nomads, but it's because it depends on sql-de-lite, so that doesn't count
17:31:40 < zbigniew> all right, ultimately, it looks like only fps and sql-de-lite are having a problem; hashes counts but has a bunch of other problems
17:32:10 < zbigniew> i assume sql-de-lite's problem is due to lru-cache
}}}"	defect	closed	critical	4.9.0	core libraries	4.7.x	fixed