« first day (812 days earlier)      last day (723 days later) » 

jordancurve
jordancurve
05:03
@ngn Notified tech@ about the "!length" error with k /tmp/in.
 
12 hours later…
coltim
coltim
17:29
hmm, what's the difference between something statically parseable versus having rules to resolve ambiguities?
ngn
ngn
@coltim static parsability is usually contrasted not with "rules to resolve ambiguities" but with the need to look up runtime information to determine the structure of the code
@coltim why are you asking? something to do with apl/j?
coltim
coltim
@ngn I guess something like {z x\y}, where when it's called like f[+;1 2 3;4] versus f["_";"a_b_c";#:]
ngn
ngn
17:44
@coltim i don't understand what you mean
coltim
coltim
well in one z is the seed for the scan and in the other it could/would be normal function application on a list of strings
ngn
ngn
@coltim no, in the second case z is still the left arg for x\
coltim
coltim
but if x is "_" there is no left arg? I guess that's the "rules to resolve ambiguities part", like `x` is treated as a scan even if x is not a function (because you can't tell without knowing what x is)
ngn
ngn
z x\y parses as x\[z;y] regardless of what x y z are
(or, more precisely: regardless of what values the variables x, y, and z have at runtime)
coltim
coltim
@ngn but if x is a char/string or an integer literal then it's parsed as z@x\y
ngn
ngn
17:53
@coltim is it really?
dzaima
dzaima
@coltim right, that's what static parsing (as it statically determines the syntactical type of each expression) means
ngn
ngn 
ngn/k, (c) 2019-2021 ngn, GNU AGPLv3. type \ for more info
 f:{z x\y}
 f["_";"a_b_c";#:]
'rnk
{z x\y}
   ^
coltim
coltim
@ngn hmm, I guess only if y is parsed as a verb (which variables aren't)
ngn
ngn
^^rank error because split (x\) can take only one arg
same thing when f is {z "_"\y}
coltim
coltim
@dzaima I guess the part I don't understand is what if you just define a b c to be b[a;c]
dzaima
dzaima
17:57
@coltim then what is f g x?
ngn
ngn
@dzaima i was about to ask the same :)
coltim
coltim
@dzaima if they're all variables then g[f;x]?
dzaima
dzaima
@coltim but, see, you have to know whether they're functions or not. Which you don't until runtime
ngn
ngn
@coltim "variables" in the sense of name-value mappings? or "not functions"?
dzaima
dzaima
static parsing just says "eh, all variables are nouns" and solves the whole problem, at the cost of making you unable to easily define custom infix functions
coltim
coltim
18:00
couldn't it just throw a run time error? 0_o (I can't tell if I'm just being dense here. I guess that would break chains of UDFs being called but maybe "forcing" it to parse as a[b[c]] would be better?)
@ngn name-value mappings
dzaima
dzaima
@coltim what should throw a runtime error?
and k already parses var1 var2 var3 as var1[var2[var3]]] always
coltim
coltim
@dzaima in the f g x example are f and g functions and x some data?
dzaima
dzaima
@coltim You don't know. That's the whole problem.
coltim
coltim
hmm, I guess f and x are less relevant. if g is a function then you could try calling g[f;x] and throw an error if that doesn't make sense. if g is data then call g[f;x] and error if that doesn't make sense?
dzaima
dzaima
but then you'll never evaluate it as f[g[x]], which is what you want most of the time.
also then -!3 would be interpreted as ![-;3]
coltim
coltim
18:21
hmm. it may just be semantics then. resolving ambiguities by parsing x y z as x[y[z]] (or noun-noun-noun I guess), or z x\y as x\[z;y] "allows" the language to be statically parseable
I've always just gotten tripped up by using \ as a function, and haven't quite puzzled out all the nuances there. at least I think I'm pretty close to understanding . =P
dzaima
dzaima
the ambiguity resolution isn't done while parsing x y z, but when parsing x (or y, or z) - x/y/z alone parse to a noun
if you replace a variable name with "abc" (or (!3), or 2, or anything else that's a noun), the program will always stay parsed exactly the same way
coltim
coltim
18:42
yeh, I'm probably just missing something. I understand that there is a grammar that defines the high-level stuff but I'm more interested in the gnarly bits like 0.1/.1/.a/a.b/a.1/,.x/(,).x/2000.01.01 and how those are handled
or rather, where the limits are with that sort of stuff, lol
dzaima
dzaima
that's tokenization then
coltim
coltim
hmm, yeh, that's probably what I get tripped on. like the result feels similar (i.e. the role something takes on varies depending on what it's surrounded by) even if the process is different

« first day (812 days earlier)      last day (723 days later) »