00:51 Manifest0 left
thowe what the idiomatic way to say "$this unless it is false, in that case nil". 01:35
I want to return NIL if it is an empty string instead of an empty string 01:36
SmokeMachine m: say “” || Nil 01:47
camelia Nil
SmokeMachine m: say “bla” || Nil
camelia bla
SmokeMachine But why do you want Nil? 01:48
thowe because it is going in to a database column that I don't want empty strings, I want NULL. Nil apparently translates to NULL. 01:51
Now, what I think I actually want to do is subclass array so that when the element is an Empty Str it returns Nil. That seems cleaner. 01:55
SmokeMachine Why not Str? 02:05
If using Red (github.com/FCO/Red), you could just not set the value… 02:08
ab5tract thowe: I’m a bit surprised at that behavior to be honest. I would expect a database driver / library to treat undefined objects as NULL on insert 02:42
thowe Empty strings don't insert as other data types. They are also less convenient to detect as lack of data. Nil inserting as NULL makes the most sense. 07:25
SmokeMachine thowe: I meant the type object Str 07:29
09:13 dakkar joined
monkeyinthejuice How do I making ARRAY of HASH 10:26
nahita3882 hi, [ %(a => 2, b => 3), %(e => 8, a => 77, f => 0) ] is an array of hashes 10:53
%(...) is one way of making a Hash literal; I'd say it's the most common and unambigous (unlike with {} maybe) 10:54
[...] is making an array
and we assign this to my @a = ... for example, and now @a is an Array 10:55
@ "sigil" is signaling that the variable is "Positional", i.e., it is a container that you can index it with natural numbers; the default class that does the Positional role is Array, therefore @a is of type Array 10:57
its cousin is List -- it also does positional but the main difference is it's immutable (its size shall never change, you cannot really assign to its elements) 11:00
and you-cannot-assign-to-elements of Lists is achieved by lack of "scalar containers" on the elements of a List unlike an Array 11:01
docs.raku.org/language/containers#...sty_things 11:02
one way of making a List is my @b is List = ... 11:05
it says it wants a Positional that is a List (not the default Array)
monkeyinthejuice How do I making ARRAY of ARRAY of hash 11:20
nahita3882 if you have Xs (X1, X2, ...) and you do [X1, X2, ...] it's an array of Xs 11:39
14:31 Manifest0 joined
rcmlz Is shrek a real person or a bot? 15:32
thowe SmokeMachine: well... I suppose one needs the ability to store an empty string if required, so silently changing that to NULL is probably not a good default(?) 15:47
ab5tract thowe: Str is not representing an empty string. It represents an undefined string 15:55
I would change the logic that produces the data to generate Str instead of an empty string 15:56
I don’t know what db library you are using but I wouldn’t personally use one that doesn’t treat an insert of Str into a string column as a NULL 15:57
thowe I'm getting the empty string by splitting a line of comma separated values. When nothing is between the commas, and using the array element to insert into the database, it is used as an empty string. 15:58
monkeyinthejuice Why does person talking begin with <letters>
@Raku bridge why
ab5tract m: my Str @a; @a[4] = “value”; dd @a
camelia Str = Array[Str].new(Str, Str, Str, Str, "value")
lizmat monkeyinthejuice otherwise on the IRC side we wouldn't be able to see who said what 16:00
ab5tract m: my @a = “,,,,”.split(“,”).map: { $_ || Str }; dd @a
camelia [Str, Str, Str, Str, Str]
ab5tract The point is that the db driver should be smart enough to go Str (or any undefined type object that matches the column type)-> NULL 16:01
thowe ab5tract: well, that may be. I'm not sure I know enough about potential issues to have a strong opinion. I think you would have to take that up with the author(s) of DB::Pg. 16:03
My script is quick and dirty, and I am using it as an excuse to learn a couple random things. A more mature version will abstract that data through validation and an intermediate data construct. in the meantime, I wanted to create a code example for myself of overloading the "[]" postcircumfix operator for an array subclass to spit out Nil instead of an undefined string, I think. 16:05
ab5tract thowe: Are you saying that you have already tried inserting undefined strings (not empty ones) using DB::Pg and the result is not a NULL value in the respective db column? 16:06
thowe I've tried whatever ends up in the array. I'm not 100% sure if they are empty or undefined. 16:07
standby
16:08 MasterDuke joined
librasteve m: enum DBCode <NULL>; sub nullify (@a) { @a.map: {$_ ?? $_ !! NULL} }; ['a','b','',Nil,Str].&nullify.map(*.raku.say); 16:08
Raku eval "a" "b" DBCode::NULL DBCode::NULL DBCode::NULL
SmokeMachine Im almost sure undefined Str will end up as NULL on Pg… I use that on Red, if I’m not completely mistaken… 16:09
ab5tract So if you want to override [], you can define an AT-POS method in a class. This will give you control of [] on scalar instances of that class ($foo[]). Storing instances in @foo requires doing the Positional role, which may be a bit more involved (haven’t done this in a while) 16:11
thowe m: dd split(',', "this,is,a,line,,,only")
camelia ("this", "is", "a", "line", "", "", "only").Seq
thowe I mean, those look like empty strings to me. 16:12
right?
librasteve ^^ if what you really want is to make "" be NULL (along with other undefined such as Nil), then testing for Truthiness does that --- personally I would want to store an empty string as "" in the DB which is different from undefined, of course
ab5tract For a quick and dirty script, my impulse would be to wrap split instead and return Str (or Nil if you really need that) instead of empty strings
thowe: so map the result like I did above and voila? 16:13
thowe the original data is not meant to be an empty Str. It is meant to be "no data". And the empty Str that results is not suitable for most of the DB columns; NULL is. 16:14
ab5tract I’ve only been saying that an Str should serve the same purpose as Nil as regards to the database insert.
And yet I have never said an empty string should result in a NULL 16:15
librasteve sorry - that's my misinterpretation of the original question --- I would defer to ab5tract and smokemachine on the best answer 16:16
ab5tract m: dd Str === “”; dd Str ~~ “”
camelia Bool::False
Bool::False
thowe Nor am I saying that. An intentionally inserted empty Str should be an empty string in the DB. The problem is I have empty Str when what I want is Nil. This data should not be empty Str in the database, it should be NULL. Therefor I need to correct the fact that they are empty Strings.
SmokeMachine I have no Pg here now... but on SQLite type objects seems to be NULL: www.irccloud.com/pastebin/8dGGoj2B/ 16:17
ab5tract m: dd split(',', "this,is,a,line,,,only").map: { $_ || Str } 16:18
camelia ("this", "is", "a", "line", Str, Str, "only").Seq
thowe OK. I have no opinion on that. If you think DB::Pg has funny behavior you will have to take that up with someone else. 16:19
SmokeMachine empty string is not Str type object... Str type object makes more sense to be inserted on DB than Nil to represent NULL
16:19 MasterDuke left
ab5tract For dealing with the kinds of frustrations that will result from trying to differentiate (‘“”,””’) from (‘,’), I think using Text::CSV is probably the best option 16:19
thowe Perhaps, but my goal at the moment was education 16:20
SmokeMachine Accepting Str as NULL is not a "funny behaviour", but the "correct" one...
thowe SmokeMachine: as I said, I have no opinion on that. You will have to talk to the PG::Db people. 16:21
er DB::Pg
ab5tract thowe:from what you have written there has been no indication that you have actually tried to insert an Str type object 16:22
thowe I'm just working with it.
According to .WHAT, the empty portions are Str
ab5tract I hope you can see how this might be frustrating to the people here trying to aid you in your education?
SmokeMachine Have you tried inserting Str instead of “”?
ab5tract Are you just ignoring every example I have posted? Because that is also frustrating 16:23
thowe m: my @line = split(',', "this,is,a,line,,,only"); @line[4].WHAT;
camelia ( no output )
SmokeMachine We are not saying objects of type Str (that is the case of “”), but the Str object type it self… 16:24
m: say Str # <- this
camelia (Str)
thowe well that doesn't help.. in the REPL that says (Str). So yes. a Str that is empty is what it is inserting. And it is landing in the DB as an empty Str. Which is what I would expect.
SmokeMachine They are 2 different things: “” and Str… 16:25
m: say “”; say Str
camelia
(Str)
thowe But I don't want empty strings. I want NULL. Inserting Nil is what works. So I want to convert those empty strings to Nil. I'm not sure what has you upset. When I split it I get "". When I look at that field with WHAT it says it is a Str. So I appear to have an empty Str, not an undefined Str. So what. converting to Nil gets me the behavior I want and the DB columns the data I want in them. 16:28
SmokeMachine m: multi to-db(Str:D $str) { $str }; multi to-db(Str:U $str) { “NULL” }; say to-db “”; say to-db Str; # thowe, this
camelia
NULL
ab5tract m: dd Str === “”; dd Str ~~ “”; dd “” ~~ Str;
camelia Bool::False
Bool::False
Bool::True
ab5tract So just put a Nil instead? You got the answer for how to do that on your first asking it. 16:29
SmokeMachine thowe: I’m not upset, I’m trying to explain to you that you are getting it wrong… you are thinking I’m saying to you to sent “” to the database, and that’s NOT what I’m saying… I’m saying to you to send Str (and not “”) to db… 16:30
thowe That's what I am doing! I was simply asking for info on the right syntax to achieve that in a couple of ways. Others tried to make it a bigger deal than that, not me. 16:31
I'm not sure what the difference is between sending "" and Sending Str is. Both appear to be what is in that data, and the result is not what I want. I am changing to Nil and getting what I want. So I'm not sure I much care what the esoteric difference between "" and Str that says it is "" is. 16:32
SmokeMachine docs.raku.org/syntax/Type%20Objects 16:33
ab5tract This is because it is not idiomatic to use Nil as the undefined value for a string. The bigger deal part is that multiple people have tried demonstrating in multiple ways what the distinction is.
SmokeMachine That’s the point… “” is a value, Str is the type itself…
thowe all I wanted to know was a good idiomatic way to convert it to Nil. The simplest thing seems to be the map previously suggested. 16:34
Yeah I get that Str is an object. And that "" is a value. The Array elem is a Str with value of "".
ab5tract m: my Str $f; dd $f; dd $f = “value”; dd $f = Nil
SmokeMachine A good idiomatic way is to NOT use Nil…
camelia Str $f = Str
Str $f = "value"
Str $f = Str
thowe But Nil does what I want, and in my head, sending Nil when I want the DB column to be NULL makes perfect sense. 16:35
SmokeMachine So does Str
thowe OK, that isn't doing what I want...
I'm not going to "fix" DB::Pg, I'm going to get the behavior I want and move on... 16:36
SmokeMachine Because you are sending “” instead of Str
thowe I'm sending @splitline[1]
SmokeMachine That’s not DB::Pg fault… if you send Str it will (probably) work!
thowe if [1] has a Str with value "" the DB thinks I am inserting an empty string. 16:37
SmokeMachine Try sending @splitline[1] || Str
I’m not saying to send Str with any value, but Str
thowe How does that make more sense than @splitline[1] || Nil ?
when what I want in the DB is NULL. 16:38
SmokeMachine Because Str is an undefined string…
thowe The DB fields are all different things, Numerics, timestamps, geographic polygons... I often want them to be NULL. Why would I send Str? 16:39
SmokeMachine Please try that and see if that works (you can change it back if you don’t like it… I’m just trying to show you that works)
thowe But I don't understand why I would care?
SmokeMachine So, it would be better to send Any…
If you don’t want to understand type objects, that’s ok… but type objects are VERY used in raku and that would help you to understand that… but if tou don’t care, I’m ok with that 16:41
ab5tract @splitline[1] || Nil will break for integer values of 0, for example
you will be sending Nil instead of 0, which may not be what you want
thowe OK, I won't be, but that's a good argument. 16:42
ab5tract also, as I tried to demonstrate a moment ago, assigning Nil to a type-constrained container results in setting that container to the type object itself 16:43
m: dd my Str $s = Nil
camelia Str $s = Str
SmokeMachine m: my Int $a is default(42); my $a = 13; say $a; $a = Nil; say $a; # thowe, be careful with this
camelia Potential difficulties:
Redeclaration of symbol '$a'.
at <tmp>:1
------> my Int $a is default(42); my $a<HERE> = 13; say $a; $a = Nil; say $a; # thowe
13
42
SmokeMachine 16:43 <SmokeMachine> m: my Int $a is default(42); $a = 13; say $a; $a = Nil; say $a; # thowe, be careful with this 16:44
m: my Int $a is default(42); $a = 13; say $a; $a = Nil; say $a; # thowe, be careful with this
camelia 13
42
SmokeMachine thowe: ^^
Nil is not NULL nor undef… 16:45
thowe I didn't think it was, I just knew that inserting it in the DB resulted in a NULL column, which was what I wanted. This is a big table with a lot of data, it will take me some time to get a test going. Right now just testing the || Str 16:50
at first bluch it might work as my error moved elsewhere to some data that is malformed... 16:52
blush
ab5tract gist.github.com/ab5tract/89c42a1eb...00e6dfceb6 17:01
this is probably how I would approach a quick'n'dirty "schema'd CSV parser"
*would start to approach 17:02
17:40 dakkar left

Powered by Cro and the Raku® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!