Interpreter bug reports in 2007 listed in the order when they were submitted. Bugs in 2005, 2006, current; library bug reports.
Contents
sparse arrays boxable (causes errors)
As reported by DevonMcCormick in the "Sparse array accumulation work-around" thread in the Forums, it is possible to end-run around the < sparse nonce error to box sparse arrays:
a =. $.&.> a: datatype a boxed datatype > a sparse boolean ] b =. $. L: 0 ;/ i. 4 4 +-----+-----+------+------+ |1 | 1|0 | 4|0 | 8|0 | 12| |2 | 2|1 | 5|1 | 9|1 | 13| |3 | 3|2 | 6|2 | 10|2 | 14| | |3 | 7|3 | 11|3 | 15| +-----+-----+------+------+
Using this type of array can result in problems:
;b scheck a rank |system error
-- DanBron 2007-01-06 17:48:57
Fixed for J6.02. -- RogerHui 2007-03-01 00:45:08
spurious errors from commuted raze-cut
As reported in the Forum's "four boxes" thread, the special code for raze-cut appears slightly broken: {{{ NB. All equivalent per Dictionary specification
f =: ([: ; <@:,;.0~) g =: ([: ; <@:,;.0 )~ h =: ([: ;@:] <@:,;.0 )~ i =: ([: 3 :'; y' <@:,;.0 )~ j =: ;@:( <@:,;.0 )~ k =: ;@:( <@:,;.0~) (i. 5 5) f 1 ,: 2 NB. Correct
5 6 7 8 9 10 11 12 13 14
- (i. 5 5) g 1 ,: 2 NB. Bug
|domain error: g | (i.5 5) g 1,:2
- (i. 5 5) h 1 ,: 2 NB. Correct
5 6 7 8 9 10 11 12 13 14
- (i. 5 5) i 1 ,: 2 NB. Correct
5 6 7 8 9 10 11 12 13 14
- (i. 5 5) j 1 ,: 2 NB. Bug
|domain error: j | (i.5 5) j 1,:2
- (i. 5 5) k 1 ,: 2 NB. Correct
5 6 7 8 9 10 11 12 13 14 }}}
-- DanBron 2007-02-08 19:39:33
Fixed for J6.02. -- RogerHui 2007-02-14 22:26:21
b. obscures -∞ ranks
The phrase u b. 0 obscures information when one or more of u's ranks are __; the adverb replaces such ranks with _ :
u"0 __ 1 b. 0 0 _ 1 9"__ i. 3 NB. Rank operates properly. 9 9 9
This behavior does not contradict the Dictionary, any more than if b. replaced (ranks of) 0 with _, so this "bug report" may be a "feature request".
While it is defensible to substitute _ for ranks which are relative (to the shapes of the arguments), a rank of __ is absolute: it is effectively 0. Only negative finite ranks are relative.
-- DanBron 2007-04-11 14:13:47
There is no bug here. See below. -- RogerHui 2007-04-12 15:03:22
+/@(]"__) b. 0 _ _ _ +/@(]"0) b. 0 0 0 0 +/@(]"__) 1 2 3 4 5 15 +/@(]"0) 1 2 3 4 5 1 2 3 4 5