list index out of bounds

[hs]

can we have it graciously not to stop the running program?

personally i like the old way of returning the last element
it's because my program input can sometimes be an empty list (i have no control of my data source)

(set 'b '())
(b 0);program stops running
(b -1);progam stops running
(0 1 b);program stops running
(-1 1 b);proram stops running

i can put a condition of (if (= b '()) (do something)
but sometimes my input can be empty, one, or two elements
so my old codes of (input 2), (input 1), etc break
yes i can put conditions too
but that makes my program long and ugly
my past few weeks were of maintenance nightmare

my main gripe is that newlisp just stops running due to list index out of bound
it's like watching your favorite dvd and your software media player (being too smart) stops running due to a scratch
while the hardware dvd player (being dumb) just keep running the dvd till the end

sorry for my rants

[cormullion]

Yes, I have some sympathy, as you might have read earlier this year (http://www.alh.net/newlisp/phpbb/viewtopic.php?t=2093, and http://www.alh.net/newlisp/phpbb/viewtopic.php?t=2084).

It's true I miss the old friendly and relaxed behaviour, but I appreciate the reasons for the sterner approach - I've rationalized the 'improvement' now as a lesson to me not to write sloppy code in the first place... :)

[hs]

yes, i do notice my code is less sloppier than it used to be
but now it consists of batallions of special cases
big, bulky code, mentally exhausting

is there any 'best practice' to avoid index out of bound cases, especially when u have less control over the list

basically i want my code to just run, not suddenly stops due to special rare case

[HPW]

Maybe you can make your own 'error-event'.
Not sure if it will work in this case.

[Lutz]

The empty list a logical false:

Code: Select all

(if lst (lst 0))

only returns the first element if not empty, else returns nil.

[hs]

i'll have to dig the error event, thx

my recent case is that i have a newlisp program in remote server to get info about time from some website, write the time to a file to be served via ajax

after regex, the time string t can be say "17:00". "17", or ""

(set 'T (parse t ":"))
;;;T can be
("17" "00") ;(T 0) (T 1) work
("17") ;(T 1) break, program stops running
(); (T 0) break, newlisp stops

my fix was to use couple if statements
the older newlisp would probably give 17:17
which is okay, because at time 17:01, the whole program (if using older newlisp) would work as nothing happens ... accuracy is not very important in my case

newlisp stops fetching and updating the ajax file, that's big problem, especially at remote server, especially when my eyes are not on that ajax file

[cormullion]

Yes, I see... I write code like that too :)

Perhaps a quick solution might be to pad out the original time string a bit.

Code: Select all

(set 'T (0 2 (append (parse t ":") '("00" "00"))))

so you'll always end up with two strings in T.

[cormullion]

Actually, though, you could just use slices rather than indexes. Slicing doesn't generate errors as readily as indexing:

Code: Select all

>(set 't "")
""
> (set 'T (parse t ":"))
()
> (0 1 T)
()
> (T 0)

list index out of bounds
>

[hs]

oh yeah i forgot to mention that i have to adjust the time to my local time
say from 17:00 to 19:02 ... so i have to get 2 values

either select or slicing returns a list, not an element
(0 1 (parse t ":")) ;gives '("17")
(1 1 (parse t ":")) ;gives '("00")

so eventually i will need to use
(set 'hr ((parse t ":") 0))
(set 'mt ((parse t ":") 1))

i keep wondering, what function to use to extract an element from a list ... (first) comes to mind

i find inconsistency for (fist) .. (by accident, just now)

(set 'b '())
(if b (b 0)) ;gives '()
(b 0) ;list index out of bounds
(first b) ;gives nil ... shouldn't it give error too?

thx

[cormullion]

i keep wondering, what function to use to extract an element from a list ... (first) comes to mind

Yes. pop will do it too, although it's destructive.

(set 'b '())
(if b (b 0)) ;gives '()
(b 0) ;list index out of bounds
(first b) ;gives nil ... shouldn't it give error too?

Yes, perhaps it should. But, at a guess, the functional nature of newLISP has overrided that: the first function has successfully analysed the list and returned nil as the answer. Perhaps it managed to find out that the list was empty without having to use index numbers... :)