Thanks Raul and Devon.
You echo my own line of thinking. So I felt reassured I wasn't missing a trick.
Then, mulling over your answers, I thought: "it's a Veeder counter!
Why not use the Veeder algorithm?" To the younger generation: a Veeder
counter is a non-electronic click-around cycle meter.
Here's the algorithm in explicit form:
bump=: 3 : 0
try.
(}:y), '0123456789' {~ >: ". {:y NB. inc last digit only
catch.
'0',~ bump }:y NB. inc (}:y) and append '0'
end.
)
Note the use of: '0123456789' {~ -in place of: ": to force an error at nnnn9.
(...Shame there isn't a noun: (n.) -: '0123456789') analogous to (a.).)
And here's my first proper go at a complex tacit verb (with heavy use
of (13 : ) and (f.) I must confess ;)
bump=: (}: , '0123456789' {~ [: >: [: ". {:) ::('0' ,~ [: $: }:)
bump y=: 'abc00098'
abc00099
bump y9=: 'abc00099'
abc00100
bump yz=: 'abc09999'
abc10000
Is there any advantage over Raul's incsfx I wondered? Mine looks
shorter, but surely such a cloddish recursion is going to be
inefficient? So I did some timings:
timer 'incsfx y'
5.00488e_5
timer 'incsfx y9'
5.41687e_5
timer 'incsfx yz'
5.38635e_5
timer 'bump y'
1.38855e_5
timer 'bump y9'
2.59399e_5
timer 'bump yz'
3.09753e_5
...Made my day!
Thanks, guys, for helping turn the hay in my mind.
I'm really sold on ($:) and (::) now... though the latter does seem
like cheating :)
On Tue, Feb 1, 2011 at 3:33 AM, Devon McCormick <[email protected]> wrote:
> I usually rely on knowledge of the prefix - like using 3&}. - but a more
> general solution might use something like this (in J7):
>
> *./\&.|.'a2b33c00100' e. Num_j_
> 0 0 0 0 0 0 1 1 1 1 1
>
> to generate the boolean selector for the number portion. This assumes that
> the prefix has a non-numeric ending to separate it from the number part.
>
> On Mon, Jan 31, 2011 at 8:03 PM, Ian Clark <[email protected]> wrote:
>
>> Smart stuff.
>>
>> I make heavy use of my own routines to do this sort of stuff and I see
>> I needn't have bothered. It's not my inability to cobble up what's
>> needed, as it's needed (a mundane skill for someone who hates looking
>> up documentation), but the time entailed in testing out homespun
>> routines (voluntary or enforced). Freedom from gotchas is more vital
>> to me than naked machine efficiency, so it's good to use something
>> venerable like printf.
>>
>> Henry's solution will suit a requirement I often encounter: handling
>> ids of the form: abc00123, ie treating those trailing digits as a sort
>> of Veeder counter. But let me just ask this, as a rider to Skip's
>> request:
>>
>> What's people's fave way to generate the next (or previous) id in a
>> series, eg 'abc00100' from the string 'abc00099'?
>>
>> Much software bilks the user, stepping say from: "image9.tiff" to:
>> "image10.tiff". This is unhelpful when sorting generated filenames in
>> a Fileman or Finder window
>>
>> Ignore the obvious problem with: abc99999 --depend on having enough 0s
>> in there. Don't demand a parameter like 3 or 'abc' to isolate the
>> suffix. Oh, and don't assume the prefix is digit-free. There can be 1
>> to 8 digits in the suffix, so you can employ 100000000+y and drop the
>> '1'. Restrict to +integer suffixes (stepping back abc00000 should
>> wraparound to abc99999) but beware the prefix might actually end with
>> '-' or '_', viz: abc_002.
>>
>>
>>
>> On Mon, Jan 31, 2011 at 10:53 PM, Skip Cave <[email protected]>
>> wrote:
>> > On 1/31/2011 3:29 PM, Henry Rich wrote:
>> >> > 'abc _%d = %s.txt' vbsprintf (3{.a) ;"0 (3{. b)
>> >
>> > That's very nice. I hadn't looked into printf much before, but it makes
>> > those kinds of text manipulations pretty easy.
>> >
>> > Thanks for the help.
>> >
>> > Skip
>> > ----------------------------------------------------------------------
>> > For information about J forums see http://www.jsoftware.com/forums.htm
>> >
>> ----------------------------------------------------------------------
>> For information about J forums see http://www.jsoftware.com/forums.htm
>>
>
>
>
> --
> Devon McCormick, CFA
> ^me^ at acm.
> org is my
> preferred e-mail
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
