I know you mentioned 8!:2. Personally, I would probably skip that and use sprintf coming from a C background...
load 'format/printf' '%d-%02d-%02dT%02d:%02d:%02d' sprintf 6!:0'' 2014-03-02T05:47:14 It's convenient that the indexes line up, otherwise it'd be a handful of selections This took less than 2 minutes, which is more than what I spent reading the 8!:2 page. When would 8!:2 be used over sprintf? Presumably 8!:2 is faster On Sun, Mar 2, 2014 at 4:04 AM, Raul Miller <rauldmil...@gmail.com> wrote: > Given > 6!:0'' > 2014 3 2 3 30 46.034 > > I'd like to generate the string > 2014-03-02T03:30:46 > > And, to make things just a bit strange, I'd like to use 8!:2 instead of the > obvious. > > Just to be clear, though, here's my idea of "obvious": > > }:;(1,5#3)<@}."_1'--T:: ',.~":,.10000+<.6!:0'' > > By obvious, I do not mean that I instantly thought of that expression, I > had to think a few minutes, and I had a few errors to fix in my initial > concept. But by "obvious" I mean that I am using the basic J vocabulary. > > Picking up the vocabulary isn't instant, though, and I'm using 14 different > words from the vocabulary (plus some nouns and a pair of parenthesis). So > if we expect a person to pick up five words from the vocabulary a week, > that's at least three weeks of training before a person could be reasonably > comfortable composing a sentence like that. > > Now... the 8!:0, 8!:1 and 8!:2 verbs do have their own tiny vocabulary as > represented by their format strings. But these are examples of extremely > specialized words - not very powerful at all, which means they require less > study than J as a whole, and (within their limited capabilities) will tend > to be faster that a more general approach. > > Formatting is not likely a speed bottleneck for iso date formatting - it's > more when you have a big sheet of numbers and a weak computer that you > might care about formatting time. But it's still an interesting exercise. > (Or a boring one, if you are not interested in formatting text.): > > So... here's how I approached this exercise: > > <.2014 3 2 3 30 46.034 > 2014 3 2 3 30 46 > > Let's just ignore the decimal part, for now. > > '4' 8!:2 <.2014 3 2 3 30 46.034 > 2014.00003.00002.00003.000030.000046.0000 > > Looks like a number by itself is decimal places after the decimal point. > > '4.' 8!:2 <.2014 3 2 3 30 46.034 > 2014 3 2 3 30 46 > > But a decimal point in the format specifier fixes that. > '4.,2.,2.,2.,2.,2.' 8!:2 <.2014 3 2 3 30 46.034 > 2014 3 2 33046 > > I need commas to separate the format specifiers for independent columns. > > '4.,r<0>2.,2.,2.,2.,2.' 8!:2 <.2014 3 2 3 30 46.034 > 201403 2 33046 > > The r option gives "fill" or, in this context "leading zeros", but each > format needs this spec. > > '4.,r<0>2.,r<0>2.,r<0>2.,r<0>2.,r<0>2.' 8!:2 <.2014 3 2 3 30 46.034 > 20140302033046 > > Getting bulky, but now all my two digit numbers can have leading zeros. I > could get fancy here, and use J to build the format specifier, but that's > premature, watch: > > '4.,p<->r<0>2.,p<->r<0>2.,p<T>r<0>2.,p<:>r<0>2.,p<:>r<0>2.' 8!:2 <.2014 > 3 2 3 30 46.034 > 2014-3-2T3**** > > Here, I've added a prefix for each number, to get the various separators. > Unfortunately, I forgot to widen each column to make things fit. > > '4.,p<->r<0>3.,p<->r<0>3.,p<T>r<0>3.,p<:>r<0>3.,p<:>r<0>3.' 8!:2 <.2014 > 3 2 3 30 46.034 > 20140-30-20T3:30:46 > > That's... better ,but now we can see that the concept of fill interacts > badly with the concept of using a prefix, so let's try using a suffix > instead: > 's<->4.,s<->r<0>3.,s<T>r<0>3.,s<:>r<0>3.,s<:>r<0>3.,r<0>3.' 8!:2 <.2014 > 3 2 3 30 46.034 > |domain error > > Oops, suffix is a q, not an s. > > 8!:2<.2014 3 2 3 30 46.034 > 'q<->4.,q<->r<0>3.,q<T>r<0>3.,q<:>r<0>3.,q<:>r<0>3.,r<0>3.' 8!:2 <.2014 > 3 2 3 30 46.034 > ****03-02T03:30:046 > > Almost done but I need to adjust my column widths again. > > 'q<->5.,q<->r<0>3.,q<T>r<0>3.,q<:>r<0>3.,q<:>r<0>3.,r<0>2.' 8!:2 <.2014 > 3 2 3 30 46.034 > 2014-03-02T03:30:46 > > Done. > > And, that's not actually a whole lot more verbose than the other expression. > > It's tempting to think about making 8!:2 more powerful, but it's not really > about power - it's a specialized tool for a specialized task. If you want > the full generality of J, it's available. > > Of course, if you really want an iso formatted timestamp, you could always > find another way to get that. Learning exercises necessarily restrict > themselves in various ways. > > Also... http://en.wikipedia.org/wiki/ISO_8601 points out that the standard > really covers a lot of ground. And one thing we do not currently have in J > is a way of getting at what the operating system thinks is the time zone of > the user. That can matter if you are on a phone (where back a few decades > ago, computers were a bit harder to carry). > > Thanks, > > -- > Raul > ---------------------------------------------------------------------- > For information about J forums see http://www.jsoftware.com/forums.htm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
